- Email: [email protected]
Heavy-section steel technology program: Intermediate-scale pressure vessel tests
Nuclear Engineering and Design 46 (1978) 47-63 O North-Holland Publishing Company
HEAVY-SECTION STEEL TECHNOLOGY PROGRAM: INTERMEDIATE-SCALE PRESSURE VESSEL TESTS * R.H. BRYAN, J.G. MERKLE, G.C. SMITH and G.D. WHITMAN Oak Ridge National Laboratory, Oak Ridge, Tennessee 3 7830, U.S.A. Received 22 September 1977 Eight flawed 990-mm-OD, 152-mm-thick vessels of ASTM A508 class 2 forging steel or A533, grade B, class 1 steel plate have been pressurized hydraulically to burst or rupture. The rupture test of one vessel (V-7) was repeated pneumatically to study the effects of the sustained load thus attainable. Test temperatures ranged from 0 to 91°C so that transitional to upper shelf fracture toughness behavior was observed. Pretest predictions of failure pressures were based on tensile and toughness properties determined from Charpy-size to 102-ram-thick specimens representative of the test vessels. Linear elastic fracture mechanics based on strain and plastic instability analyses were generally adequate for determining the failure pressures, which ranged from 2.15 to 3.28 times design pressure.
ceived to provide conditions o f size and state o f stress similar to those that would obtain in real reactor vessels in normal operation and under accidental conditions. Although ITVs are intermediate in size between laboratory specimens and reactor vessels, they are nearly full size with respect to thickness and are capable o f providing conditions of transverse restraint not attainable by other means. The ITV tests were planned to provide a connection between the fracture behavior observed in smallscale tests and the behavior o f full-scale components. When the first two ITVs were tested it was not known whether and within what limits linear elastic fracture mechanics (LEFM) would be a useful analytical method. The general plan for the series was to observe flaw behavior quantitatively under a variety o f fracture toughness conditions ranging from low transitional to upper shelf toughness. Some consideration in the early order o f tests was given to demonstrating qualitatively that a margin o f safety in flawed vessels does exist. All o f the test conditions were selected to be related to some normal or hypothetical abnormal phase o f nuclear power plant operation. However, experimental considerations generally dictated the
1. Introduction Development and validation o f methods o f analyzing full-scale light-watercooled reactor pressure vessels containing flaws is a major objective o f the heavy-section stell technology program. Such methods are an essential tool for the evaluation o f the safety o f nuclear power plants. It is especially desirable that estimates o f the strength o f flawed structures be made realistically and quantitatively to ensure adequate margins o f safety in the interest o f public safety and to avoid unnecessary conservatism in the interest of economy and productivity. The testing o f intentionally flawed steel vessels is a necessary part o f this development effort. The interme. diate-test-vessel (ITV) series o f experiments was con* Work done at Oak Ridge National Laboratory, operated by Union Carbide Corporation for the Energy Research and Development Administration. This work funded by U.S. Nuclear Regulatory Commission under Interagency Agreements 40-551-75 and 40-552-75. Expanded version of Invited Paper G1/I* presented at the 4th International Conference on Structural Mechanics in Reactor Technology, San Francisco, California, 15-19 August 1977. 47
R.H. Bryan et aL / Intermediate-Scale pressure vessel tests
48
flaw size; and most tests were carried to the ultimate capacity of the testing system or to the point of structural failure. Extremes of temperature and pressure not directly related to any reactor operating condition were used as necessary to obtain experimental results by which methods of analysis could be decisively evaluated. The tests performed to data have been primarily concerned with the study of the onset of flaw instability [ I - 5 ] . Any method of analysis that is successful in predicting the threshold of instability must be able to relate the geometry of the flaw, structure, and loading to measurable material properties. This usually involves consideration of more than a single fracture criterion. To the extent practicable all of the variables that might have a bearing on flaw behavior were measured during each test: strains near and far from the flaw, crack-opening displacement (COD), crack depth, temperature, and pressure. In addition to the quantitative results of the tests some important conclusions regarding modes of behavior were reached. Conditions for stable crack growth, mixed mode crack propagation, and leak without break were observed.
2. Description of intermediate test vessels Ten 152-mm-thick vessels were fabricated to the configurations shown in fig. l. The 990-mm-OD by 1370-mm-long test sections were fabricated of ASTM A508, class 2 forgings in vessels V-1 through V-6 and of ASTM A533, grade B, class 1 plate in vessels V-7 through V-10. The nozzles in V-5, -9, and -10 were fabricated of ASTM A508, class 2 forging steel. These characteristics are summarized in table 1. The test cylinders of forged vessels V-3, -4, and -6 were cut longitudinally and rejoined by submerged-arc welding to provide weld metal sites for emplacement of flaws as shown in fig. 2. The test section of vessel V-6 also contained an intersecting circumferential weld. The fabrication of the vessels has been described in detail by Childress [6,7]. The vessels were designed and fabricated in accordance with section III of the 1968 edition of the ASME boiler and pressure vessel code. The thickness of the test section was chosen as the minimum adequate for the performance of valid tests of linear elastic fracture mechanics. The length was such that a central flawed region would be unaffected by the dis-
. 380 1D
I
I
I
f
r I- 99o og |
I ~ J'~--152SHELL /II i 1I Jt 123 I "--~'\TEST13170 I J~23~i 10./~ J L-
VESSELS1, 2, 3, 4, 6, AND 7
1
DIMENSIONSIN m
VESSELS5 AND 9
Fig. 1. Intermediate test vessel typical cross sections.
52WALL
49
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
Table 1 Materials used for fabricating intermediate test vessel cylinders and nozzles. Vessel designation
Material Cylindrical shell
V-1 V-2 V-3 V-4 V-5 V-6 V-7 V-8 V-9 V-10
A508 class 2 A508 class 2 A508 class 2 A508 class 2 A508 class 2 A508 class 2 A533-B class 1 A533-B class 1 A533-B class 1 A533-B class 1
Nozzle
A508 class 2
A508 class 2 A508 class 2
continuities at the two heads. The diameter of the test section and closure forging were a practical minimum allowing personnel access for inspection, flaw preparation, and instrumentation. The ITV head and closure are designed in accordance with section III of the code for a design pressure of 138 MPa. In addition, the allowable stress intensity for secondary
membrane plus bending in these components was arbitrarily set at 1.1. times the stress intensity that would obtain in an infinite cylinder at a pressure o f 138 MPa. These measures were taken to ensure that initial failure would take place in the test section. Design parameters of the test vessels are compared with those of typical boiling-water reactor (BWR) and pressurized-water reactor (PWR) pressure vessels in table 2. The ITV thickness is almost full scale with respect to real reactor vessels. Although the ITV design pressure of 66.9 MPa is markedly different from that of the reactor vessels the stress states at design pressure are similar; by virtue of the code definition of allowable stress intensity, the difference between the largest and smallest principal stresses is the same in the cylindrical sections of both types of vessels under pressure loading.
3. Test preparations Intermediate vessel tests were preceded by intensive preparations to obtain a quantitative basis for predicting failure conditions and for evaluating meth-
9e° ee 00~
99° ~'~"O0~ • ONGI,UD,NAL
\
I LONGITOOlNAL SUBMERGED-
~
,.c
1370 mm
685, mm[
J
1370 mm
iii LOCATION
GIRTH SUBMERGEDARC WELD
LOCATION LONGITUDINAL SUBMERGEDARC WELD J
(a)
(b)
Fig. 2. Weld orientations for flaw location. (a) vessels 3 and 4; (b) vessel 6.
50
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
Table 2 Nominal design parameters of cylindrical shells of intermediate test vessels and BWR and PWR vessels. Vessel
Inside Diameter (ram)
ITV BWR PWR
Thickness (mm)
686 6100 4370
Design pressure a fMPa)
152 164 216
Average stresses at design pressure fMPa)
66.9 9.64 17.3
Circumferential
Axial
151 179 175
61.7 87.3
83.6
a Code allowable for a cylinder as determined by the general primary membrane stress intensity.
ods o f fracture analysis. Material properties of the test vessels were measured, and flawed epoxy and steel models were tested. Tentative test conditions were evaluated by experimental and mathematical analysis to determine the specific test conditions, in terms of flaw size and vessel temperature, likely to produce useful intermediate vessel test results.
O 540 A I "1 20 ~
V 1 V-2 V-?
A
V
/,,
f
,,
- Y~- . . . . . . . .
I
Tensile, impact, and fracture toughness properties of the ITVs were determined prior to testing. Prolongations of the test cylinders and nozzle forgings were acquired for this purpose. Thus the material properties were determined for pieces that had experienced the same fabrication history as the vessel test sections up to the time of welding of vessel components together. Prolongations of the forged cylinders with longitudinal welds (V-3, V-4, and V-6) were separated from the vessel test sections after those welds were stress relieved. The accuracy of structural and fracture analysis is
TENSILE TEST TEMPERATURE ROOM ROOM 933°C
~
5~'
4. Material characterization
~
/
---
/o
600
500
o
'
o
/
O
¢
----
©
f
J
I
J
~
°-'
/
f
~460
4oo Lu
(
/
~ 300
~7 V
200
V
100
400
50 100 DISTANCEFROMOUTSIDESURFACEOF CYLINDER(ram)
150
Fig. 3. Lower yield stress as a function of thickness from prolongations of vessels V-l, V-2, and V-7.
0
0
2
4 6 TRUE STRAIN I%)
8
10
Fig. 4. Typical uniaxial stress-strain curve from prolongation of vessel V-1.
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
51
Table 3 Selected values of Klc d used in pre-test analysis. Vessel No.
Flawed material Test Temp. ¢C)
1 2 3 4
54 0 54 24
5 6
88 88
7 7A 7B
91 88 87
9
24
A508 A508 SA weld SA weld A508 A508 SA weld A508 A533 A533 Repair weld HAZ A508
Range of Values (MN/m3/2) near test temperature
Pre-test analysis value adjusted for section thickness (MN/m3/2)
PCCV
CT a
163-190 157-204 182-202 111-202 181-220 181-255 163-190 192-235 162-220 162-220 220 307 184-223
222-335 182-230 198-315 147-214 287-314 165-264 202-353 229-377 245-447 245-447
342 202 357 220 b 302 265 405 429 331 331
159-298
159-298
a From Mager [11]. b Post-test value 176. limited by the quality of available material property data. The variation of properties with depth were measured where practible. Yield stress data, which are especially important in tests that go well beyond yield, are shown in fig. 3 as a function of thickness for three cylindrical test sections. Except for the case of V-2, which was tested at 0°C, the tensile properties were determined near the respective vessel test temperatures. Fig. 4 shows a typical true stress-true strain relationship for V-1 material. The perfectly plastic and strain-hardening characteristics shown in this figure were important factors in analyzing the vessels. Charpy impact energy curves were determined as a function of depth and orientation. Results were typical of A508, class 2, and A533, grade B, class 1 materials. NDT temperatures determined by dropweight tests of shell course material were typically - 1 2 ° C and - 5 1 °C, respectively, for the two materials. Fracture toughness was estimated from test data from bend tests o f precracked Charpy-V (PCCV) specimens and from tests of 21.6- and 76- or 102mm-thick compact tension (CT) specimens. The 76and 102-mm CT specimens were the largest that could be cut from nozzle and shell prolongations.
Data were converted to lower bound Kicd values by the equivalent energy procedure proposed by Witt and Mager [8]. Such calculations have been shown by Merkle and Corten [9] to agree with J-integral calculations, for the same chosen measurement point, for these two specimens. The thickness of the smallest CT specimens corresponds to the thickness of 1/7scale steel model vessels used in this program. Variations of toughness with depth in cylinder wall and temperature were determined for the material in which the flaw would be located. Variability of KIcd values was substantial, as had been expected from earlier studies by Mager [ 10]. Selected properties used in pretest analyses are shown in table 3.
5. Model testing Scale models were used extensively as an aid to planning and interpreting results of tests of the ITVs. Epoxy models, which behave elastically, as well as steel models, which exhibit elastic-plastic behavior, were tested. In each case the model contained machined notches that were sharpened either by fatigue (in epoxy) or by hydrogen-charge cracking of an electron-beam weld placed at the root of the
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
52
/
CIRCULAR
\A " ~AT'~UEOW \~
//"-"~\M.ACHINED ~
IV~
I / / / / / / /,1 ?-\~Af"!~EOA l/ / / / / / / /'L/FA'TI(SUED/ / I 1//// / / I1//// //A
/',~,
\ "~\\\\\1
VESSELS 1, 2, 3, 4, AND 6 I~
2b VESSELS 5 AND 9 _
'Y//]
°.c.,..°
c..c.,° 111
l
I
I
I
I
t'/////h
~Zb °
VESSELS 7, 7A, AND 7B
Fig. 5. Flaw geometries for intermediate test vessels.
Table 4 Flaw design for intermediate-size vessels Vessel
Flaw
Location a
Material
Dimensions (mm) Estimated a
Ab B
Outside Outside Outside Outside Outside Inside nozzle
A508-2 A508-2 SA Weld SA Weld A508-2 A508-2
Actual 2b
a
2b
64 64 65 66 66 30
206 206 210 210 204
65 64 54 75 79
210 211 216 210 204
47 44 45 135 135 140 30
129 132 135 472 472 472
47 34 49
133 132 135
139
467
corner
Ab B
C 7 7A 7B 9
Outside Outside Inside Outside Outside Outside Inside nozzle corner
SA Weld A508-2 SA Weld A533-B1 A533-B1 Sect. XlWeld HAZ A508-2
a Plane of flaw lies in radial-axial plane of test cylinder (and also of nozzle in V-5 and V-9). b Flaw at which fracture occurred.
53
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
machined notch (in steel). The models were instrumented to provide pressure-strain data up to the time of failure. These data provided a means of qualifying various computational methods for determining the strength of flawed vessels as a function of flaw geometry and toughness. Shape factors needed for application of LEFM were measured in epoxy model tests.
6. Test vessel preparation Sharp flaws of the desired shape and location were necessary to satisfy the assumptions of LEFM and to maximize the chances of successful correlation of experiment and analysis. Flaw geometries for the tests are defined by fig. 5; and the estimated and actual flaw dimensions are given in table 4. In all cases the actual initial flaw geometry was close enough to the estimated value to meet test objectives; however, the discrepancies were in some instances large enough to affect the strength of the vessel and accuracy of pretest analysis. Flaw preparation involves two steps: (1) a notch is machined at the desired location to within a few millimeters of the desired final crack tip contour; (2) the machined notch is extended either by fatigue or by a combination of electron-beam (EB) welding and hydrogen-charge cracking. Fatiguing the crack by cyclic pressurization of the notch surface was accomplished satisfactorily for all ITV flaws except those of the VV-7 type. In the V-7 vessel the volume of the machined notch was so large that it was impractical to pressurize it cyclically; furthermore, uniform fatigue crack growth could not be assured around the periphery of a long deep flaw. Therefore, in V-7, V-7A, and V-7B, the machined notch was sharpened by EB welding the periphery to a uniform depth and subsequently hydrogen-charging the brittle weld until a sharp crack was formed. Flaw sharpening in each case extended the machined notch generally by 8 to 25 mm to a smooth final crack contour. Vessel V-7 has been tested three times. This was considered feasible since the test section did not yield through the thickness in regions remote from the flaw. Preparation for the last two tests of V-7, designated 7A and 7B, involved repair of the previous flaw and rupture zone. In each case the cracked region was cut out and the vessel was repaired with-
out post-weld stress relief by the half-bead welding procedure prescribed in section XI of the ASME boiler and pressure vessel code [ 12]. Measurements and studies by Smith [ 13] generally indicate that residual stresses resulting from the section XI welding procedure are low in the weld metal and approach yield in the base metal near the weld. The vessels were extensively instrumented with internal and external strain gages (typically 60 or more), crack-opening-displacement (COD) gages, ultrasonic transducers for monitoring the location of the crack front, acoustic emission sensors, and thermocouples.
7. Test description The intermediate vessel tests were performed under the conditions indicated in table 3. Prior to pressurization, the test vessel was brought to the prescribed temperature range. With one exception, the vessels were pressurized hydraulically until a leak or burst developed. Usually pressure was raised in steps of 3 to 30 MPa, depending on the pressure, with pauses in pumping to allow for recording and assessment of data. Vessel V-7A was pressurized pneumatically in an attempt to replicate the hydraulic test of V-7. The pressurization schedules for the hydraulic tests were generally short, except for the first test, in which trouble with equipment was encountered. Fig.
3o351
240 220 2OO 180 ~. 160 v
o2o I
/I.
i
. . . . . . . . . . .
15
i
0
1
! l
2
=
120
~
100 i
io ? .................
140
3
--
80
-6o
0
Time ( h r )
Fig. 6. Pressure-time record for hydraulic pressurization of ITV-3.
R.H. Bryan et aL / Intermediate-Scale pressure vessel tests
54 200
. . . .
r
. . . .
I
. . . .
I
. . . .
I
. . . .
I
. . . .
I
.
.
.
.
.
.
.
.
I
. . . .
I
. . . .
150
PE
30
IO0
5O
/
/
/1
0
t -10
0
. . . .
d 10
. . . .
J 20
,
,
,
,
I
,
,
,
,
30
l
L
,
40 TIME
,
,
I 50
,
,
,
i
,
60
,
,
,
,
I 70
I J i i J 80
90
(hr)
PRESSURIZATION SCHEDULE FOR V - 7 A TEST
Fig. 7. Pressure-time record for pneumatic pressurization of ITV-7A.
6 shows a typical record of a hydraulic test. The pneumatic test schedule was deliberately protracted because of a combination of test requirements and test facility limitations. Several loading-unloading cycles were desired and four were accomplished as shown in fig. 7.
8. Discussion of results The intermediate vessel tests demonstrated that, large flaws notwithstanding, all of the vessels were capable of yielding in the unflawed region of the test cylinder before the onset of fracture. A long, homogeneous cylinder of the diameter and thickness of the test vessels would behave as shown in fig. 8, which is a calculated pressure-strain curve based upon a trilinear approximation to the stress-strain relationship given in fig. 4 with a yield stress of 520 MPa. One can see in fig. 8 that initial yielding occurs, on the inside surface, when p = 0.26 Oy. The cylinder first yields on the outside surface at the gross yield point, where p = 0.37 Oy. The test results shown in fig. 9 for the cylindrical test sections and in fig. 10 for the nozzles illustrate the significance of the strength demonstration. Initial
yielding on the inside surface corresponds to 0.1% outside surface strain on these pressure-strain plots, a condition all vessels attained. Initial yielding occurs when the pressure reaches 2 X design pressure, the latter being 66.9 MPa. Every flawed vessel that burst sustained a pressure of at least 2.7 X design pressure without fracture, and outside surface strains were 0.168% or greater. In the tests of vessels V-5, V-7, and V-7A the vessels leaked but did not burst. Vessel V-7B did not burst, but the preliminary indications from this most recent test are that the flaw was tearing unstably at the maximum test pressure. One of the specific purposes of testing thick vessels was to study the applicability of methods of fracture analysis to the prediction of failure pressure. The limits of validity of linear elastic fracture mechanics (LEFM) was of particular interest. The three types of flaws shown in fig. 5 present three types of problems. The cylindrical vessels with part-circular flaws provided the simplest geometry for analysis and also provide conditions approaching plane strain at the deepest part of the flaw. Tests of flaws of this type were performed first. Nozzle corner flaws are in a region in which both stress concentrations and transverse restraint complicate the analysis. The deep, long-flaw geometry of vessels V-7, -7A, and -7B,
55
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
220
180
I |
/~I
Fully strain hardened-~
which was originally conceived for a specific study o f leak-without-burst, also presents a different type o f problem. The fracture analysis and test results are summarized in table 5, which compares LEFM analyses with test results. LEFM based on strain was generally used by Merkle [2] in the analyses o f the vessels. In this method fracture mechanics is used to compute a failure pressure and strain, Xf, strictly on the basis of elastic behavior. Since this failure pressure is unrealistically high for strains above yield, the estimated real failure pressure is determined as that pressure at which the computed failure strain, Xf, would obtain in an unflawed cylinder. A calculated or measured pressure-strain curve for an unflawed cylinder, as shown in fig. 8, was used to obtain the failure pressure. The results o f this procedure are in good agreement with the two tests, V-2 and V-4, in the transition range o f toughness. These two vessels fractured extensively prior to gross section yielding. Vessels V-1 and V-3 were tested at a temperature (54°C) for which the static fracture toughness was
%,-Onset of strain . . . . . k-Gross yield hardening
160
140
j f - - I ni tia] yielding
~ 12o . ~
.
.
--
i
.
--
~ ! , 1
10o ~-
'
t
I
80
'
6o
1
I
"
"
"
-. . . . . . .]-
4O 20 - -
]
G 0
0.2
0.4
0.6
0.8
1.0
1.2
Outside s u r f a c e c i r c u m f e r e n t i a l s t r a i n
1.4 (%)
Fig. 8. Typical pressure-outside circumferential strain curve for a long unflawed cylinder with intermediate test vessel dimensions.
240
220
200
~~__~-"-'--'-'~ /~.-V-1
~180
uJ 160 O: o.
140
f
V-7, V-7A
120
100
0
0.5
1.0 STRAIN (%)
1.5
2.0
Fig. 9. Outside circumferential strain remote from flaws. Cylindrical intermediate test vessels 1, 2, 3, 6, 7, and 7A. Vessel V-4 responded approximately as V-2.
56
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests 200
I
I
I
180
160
140
120
100
80
~
•
GAGE 89. V S
--
/k GAGE 9Y, V - 9
X 40
L
20
-
o 0
I
I
I
05
,0
,s
zo
STRAIN 1%1
Fig. 10. Outside ckcumferentia] strain remote from flaw. Intermediate test vesselsV-5 and V-9 with nozzles.
known to be high (above 300 MN/m 3/2) but the crack arrest toughness low. Hence crack arrest was not expected. The cracks propagated in a mixed mode fracture beyond the ends of the test sections. Substantial stable crack growth prior to fracture was expected and did occur. LEFM calculations of the critical pressure and strain, taking stable crack growth into account, were quite conservative. The gross section strains at fracture were well beyond yield. Fracture appeared to be related to the attainment of local plastic instability, which was, in turn, dependent upon the extent of stable crack growth. The latter phenomenon must be accounted for quantitatively in an analysis if the method is to be useful for predicting fracture strength after through-the-thickness yielding OCCURS.
Following the test of vessel V-I, the fracture surfaces were examined by scanning electron microscopy, under the direction of Canonico, to determine the fracture mode. Fig. 11 shows the fractured region of the vessel immediately after the test. Figs.
12 and 13 show fracture surfaces removed from the vessel and the essential results of the fractographic study. The region of stable crack growth extends from the sharpened flaw through the midsbction of the wall for some undetermined distance within the dimple-fracture region shown in fig. 12. Beyond the dimpled region, which was about 500 mm long, the central section exhibited cleavage, as shown in fig. 13. The test of vessel V-6 at 88°C gives, further confirmation to the conclusions derived from the V-1 and V-3 tests with respect to crack initiation. At the higher temperature, however, the V-6 fracture was a full shear terminating within the test section. The decreasing load during fracture probably limited the length of the tear. These five cylindrical vessel tests provided an opportunity to compare fracture behavior in two different materials as well as in different toughness regimes. The flaws that failed in V-3, V-4, and V-6 were located in submerged arc weld metal, while the V-1 and V-2 flaws were in A508 tbrgings. The flaws in weld metal behaved essentially the same as those in base metal, but prediction of failure was affected by the greater apparent variability in toughness of weld metal relative to base metal. The preparations for testing the vessels with nozzles, V-5 and V-9, included an elastic three-dimensional finite-element analysis by Krishnamurthy [14] and experimental determination in epoxy model tests by Derby [1,4] of stress concentrations at the nozzle corner and shape factors for nozzle corner flaws. The elastic stress-concentration factors obtained from the finite-element analysis and the epoxy models (2.98 and 2.79 respectively) were substantially lower than that measured in the intermediate vessel tests (~4). The preliminary shape factor deternfinations were deduced from burst tests of flawed epoxy models, for which material Kic was independently measured [4]. Recent frozen stress photo-elastic experiments by Smith et al. [ 15] on identical models agree with Derby's results. As indicated in table 5, the test pressure of the vessels with nozzles, V-5 and V-9, exceeded the fracture pressures predicted by LEFM based on strain by a considerable margin. Although the stress is high at the nozzle corner on account of stress concentration, the transverse restraint on the flaw is low. Nozzle corner circumferential strains reached about 6.5 and
57
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
Table 5 Calculated and test results Vessel No.
Calculated results
Test results
Local plastic LEFM instability pressure (MPa) Failure pressure (/VlPa)
Mode of failure
2 4 9
189 181 93-127
1 3
206 c 217 c
5 6 7 7A
232 c 132 132
7B
132
190 190 132 190 143 d 143 d 183 e 143 d 183 e
Maximum Maximum pressure (MPa) strain (%) a
Load factor b
Failure strain (%)
Transition range thoughness 0.206 Flat 192 0.163 Mixed 183 0.070-0.096 Mixed 185 Static upper shelf toughness 0.345 Mixed 199 0.398 Mixed 214 Static and dynamic upper shelf toughness 0.100 Leak 183 0.479 Shear 220 0.109 Leak 147 0.109 Leak 144 0.109
Leak with continued slow tearing
152
0.19 0.168 1.05
2.87 2.73 2.77
0.92 1.45
2.96 3.19
0.25 2.0 0.12 0.110
2.74 3.28 2.20 2.16
0.115
2.26
a Outside circumferential strain remote from flaw. b Ratio of failure to design pressure. c Assuming 15% stable crack growth. d Ligament rupture pressure. e Burst pressure.
8.5% in V-5 and V-9, respectively, in comparison with predicted strains of 0.35 and 0,6 to 1.1%. Although pretest analyses underestimated fracture pressures and strains in the vessels with nozzles the test results were predicted qualitatively. Vessel V-5, which was tested at 88°C, did not fracture; the crack grew stably with increasing pressure until a leak developed. Fracture pressure is presumably higher than the failure pressure observed in V-5. Vessel V-9 fractured at about the pressure at which V-5 leaked. Because of the low V-9 test temperature (24°C), the arrest toughness was below the static thoughness, and the crack propagated rapidly. The crack had, however, grown stably from a depth of 30 mm to about 50 mm before becoming unstable. Work on nozzle corner cracks is continuing and one additional ITV with a nozzle will be tested. The test of an intermediate vessel with a long,
deep flaw (fig. 5) was initially conceived as a demonstration of leak-without-burst. Accordingly, the size of the flaw in vessel V-7 was chosen so as to attain failure between design pressure (66.9 MPa) and gross yield pressure. Tests of flawed steel models confirmed the latter to be about 190 MPa. Pretest analyses of the two possible modes of failure were made. The leak-without-burst mode of failure obtains if the ligament at the deepest part of the flaw ruptures prior to obtaining a condition at which the flaw is unstable with respect to axial propagation. The predicted conditions for leak shown in table 5 were based upon LEFM with a ligament-size-dependent fracture toughness. This analysis also agreed with model test results. Burst estimates were based upon a LEFM analysis of a through-the-wall crack with corrections made for plastic zone size and bulging. Without considering the
c~
~J
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
ITV Vl ASTM A508 CL 2
~ i!~i~iii~i
Fig. 12. Vessel V-1 fracture surface in region of prepared flaw. The macro photograph shows the machined notch surface, the fatigued crack extension, and the region of ductile fracture. The scanning electron microscope photomicrographs show dimple failure.
59
60
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
ITV V1 ASTM A508 CL 2
Fig. 13. Vessel V-1 fracture surface in region of prepared flaw. The inset shows a region near the termination of the ductile zone which was examined by scanning electron microscopy. The photomicrographs show the transition from dimple to cleavage mode of failure.
elevation o f fracture toughness from lack of transverse restraint the analysis predicted a burst. With an adjustment for less than full transverse restraint according to an empirical relationship o f Irwin [ 16] the analysis indicated that the vessel would attain gross yield prior to burst. It was also estimated that a local plastic instability around the flaw would
develop prior to ligament rupture or burst. The V-7 test was carried out hydraulically with the expected result, a leak at 147.2 MPa. Upon depressurization the vessel resealed itself and held a pressure o f about 125 MPa. The vessel was repaired and retested pneumatically with an identical flaw in a virgin region o f the vessel. A thin-membrane patch was welded to
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
61
Fig. 14. Fracture surfaces from vessel V-7A. The light areas at the sloping ends are surfaces fractured in liquid nitrogen after the test. the inside surface beneath the flaw so as to allow the pressure to be maintained long after rupture. The sustained load test, designated V-7A, produced flaw behavior practically identical to that o f the first test. Rupture occurred at 144.2 MPa, the crack had grown about 10 mm toward each end but did not propagate. The fracture surfaces in V-7A are shown in fig. 14. The light areas at the sloping ends of the flaw were fractured in liquid nitrogen after the test. Fractographic examination shows the areas of crack-extension to be dimple fracture. In both of these tests the location of the edge of the crack was observed ultrasonically. The crack grew stably with increasing pressure, as shown in fig. 15 for the V-7 test. Stability during pressurization was confirmed by observing crack-opening displacement (COD), o f which fig. 16 shows typical behavior and a comparison of V-7 and V-7A. This indicates that COD continues to increase to some degree during periods of constant or slightly decreasing pressure, a phenomenon that was observed in all of the intermediate vessel tests. This creeping
(in.)
(rnm}
1.2
1.0
0.8
0.6
0.4
0.2
0 (MPa)
34,5
~.9
1Q~.4
I
I
I
I
15,1000
20,000
,6000
10,000 PRESSURE
137,9
Fig. 15. Crack depth and inside surface deformation versus pressure measured ultrasonically in vessel V-7.
I
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
62 (mrn)
(in) 06
15
} 0.5
uJ 10
I
V-7
V-7A GAGEZT 129 CYCLE 2 • CYCLE3 CYCLE4
0.4
v) C3
03
5 c,
~s 5
0.2
/
!
•
t
J
01
(psi)
(MPal
0 6000
I
10,000
I
50
14,000
18,000
I
22,000
Ti
100 PRESSURE
150
Fig. 16. Crack-opening displacement versus pressure for vessels V-7 and V-TA,
behavior, which had been suspected as possibly indicating incipient instability, was carefully evaluated in the test of V-7A, which was pressurized pneumatically and specially instrumented for this purpose. The stability of the crack at peak pressure is indicated by the asymptotic behavior of the COD at maximum pressure shown in fig. 17. The maximum pressure was maintained within 1.1% after rupture occurred. The third test of this vessel, called V-7B after the second weld repair, was performed by hydraulic pressurization. The flaw geometry was identical to that in the V-7 and V-7A tests. The sharpened flaw in V-7B was placed as nearly as possible within the heataffected zone resulting from the repair. A pressureretaining patch was also installed in this vessel so that the sustained-load behavior o f the flaw in the neighb orhood of the weld repair could be compared with the behavior observed in the previous tests. The vessel ruptured at 152 MPa with outside surface strains remote from the flaw very nearly the same as in V-7 and V-7A. The patch allowed the pressure to be maintained almost constant for more than
a minute after rupture was observed. At the time of rupture and during the period of sustained loading thereafter, the COD increased much more rapidly than at lower pressures and continued to do so until the patch ruptured and the vessel depressurized. Present interpretations of the behavior of V-7B are 4.6
,..,.,..r,-,
...r ,.¸ , . - .
-,-..,
......... r'--'"',"-,--,..-.,..-,
-.,..-,--.
-,....,--,-..,
....
4.4 4.2
f
v~4"O~
/
~3.8
/
3.6
~ T l m e ' of rupture
/ 3.4 / 3.2
i
400
800 1200 Time( sec )
i600
2000
Fig. 17. Crack-opening displacement versus time for vessel V-7A during final period of sustained load.
400
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
based upon preliminary evaluations of the data and incomplete examination of the flaw. In V-7B a short tear through the ligament is visible, whereas previously the rupture could not be seen until the flaw was cut from the vessel. In V-7B crack extension axially is visible for at least 50 mm at both ends, and there are preliminary ultrasonic indications that the crack runs out o f the original plane o f the flaw. The data suggest that the vessel was slowly tearing just prior to depressurization. It has not yei been determined how far the crack had grown before the COD became unstable. However, the pretest burst analysis shows that a through-the-wall crack ! 80 mm longer than the initial flaw would be unstable at 152 MPa.
9. Summary and conclusions The tests of intermediate-size vessels with sharp flaws permitted the comparison of experimentally observed behavior with analytical predictions o f the behavior o f flawed pressure vessels. Fracture strains estimated by LEFM were accurate in the cases in which the flaws resided in regions of high transverse restraint and the fracture toughness was sufficiently low for unstable fracture to occur prior to yielding through the vessel wall (viz V-2 and V-4). When both of these conditions were not present, unstable fracture did occur, always preceded by stable crack growth; and the cylinders with flaws initially less than halfway through the wall attained gross yield prior to burst (viz V-l, V-3, and V-6). Predictions of failure pressure of the vessels with flawed nozzles, based upon LEFM estimates of failure strain, were very conservative. LEFM calculations of critical load were based upon small-specimen fracture toughness test data. Whenever gross yielding preceded failure, the actual strains achieved were considerably greater than the estimated strains at failure based on LEFM (viz V-l, V.3, V-6, and V-9). In such cases the strength o f the vessel may be no longer dependent upon plane-strain fracture toughness but upon the capacity of the cracked section to carry the imposed load stably in the plastic range. Stable crack growth, which has not been predictable quantitatively, is an important factor in elastic-plastic analysis of strength. The ability of the flawed vessels to attain gross yield in unflawed sections has important qualitative
63
implications on pressure vessel safety margins. The gross yield condition occurs in light-water-reactor pressure vessels at about 2 × design pressure. This loading condition is generally well beyond the allowable normal, upset, and emergency conditions for which reactor pressure vessels may be designed. The intermediate vessel tests that demonstrated a capacity for exceeding this load (vis all except V-2, V-4, and V-7) confirm that the presumed margin of safety is not diminished by the presence of flaws of substantial size, provided that material properties are adequate. References [ 1] R.W. Derby et al., USAEC Report ORNL-4895, Oak Ridge National Laboratory, February 1974. [2] R.H. Bryan et al., USAEC Report ORNL-5059, Oak Ridge National Laboratory, November 1975. [3] J.G. Merkle et al., ERDA Report ORNL/NUREG-1, Oak Ridge National Laboratory, August 1976. [4] J.G. Merkle et al., ERDA Report ORNL/NUREG-7, Oak Ridge National Laboratory, August 1977. [5 ] R.H. Bryan et al., ERDA Report ORNL/NUREG-9, Oak Ridge National Laboratory, to be published. [6 ] C.E. Childress, USAEC Report ORNL-TM-4351, Oak Ridge National Laboratory, October 1973. [7] C.E. Childress, Documentary Report 5, USAEC Report ORNL-TM-5074, Oak Ridge National Laboratory, January 1976. [8] F.J. Witt, T.R. Mager, USAEC Report ORNL-TM-3894, Oak Ridge National Laboratory, October 1972. [9] J.G. Merkle, H.T. Corten, J. Pressure Vessel Tech. ASME, 96 J. (1974) 286. [10] T.R. Mager, Report No. WCAP-7578, Westinghouse Electric Corporation, PWR Systems Division, Pittsburgh, Pennsylvania, October 1970. [11] T.R. Mager, S.E. Yanichko, L.R. Singer, Report No. WCAP-8456, Westinghouse Electric Corporation, Nuclear Energy Systems, Pittsburgh, Pennsylvania, December 1974. [12] "Rules for Inservice Inspection of Nuclear Power Plant Components," Section XI, Subsubarticle IWB-4420, 1974 edition with Summer and Winter 1974 Addenda, American Society of Mechanical Engineers, New York. [13] G.C. Smith, ERDA Report ORNL/NUREG/TM-94, Oak Ridge National Laboratory, pp. 39, April 1977. [14] N. Krishnamurthy, USAEC Report ORNL-TM-3315, Oak Ridge National Laboratory, March 24, 1971. [15] C.W. Smith, M. Jolles, W.H. Peters, VPI-E-76-25 (also ORNL/Sub/7015-1) Virginia Polytechnic Institute and State University, November 1976. [16] G.R. Irwin, et al., USAEC Report ORNL-NSIC-21, Oak Ridge National Laboratory, December 1967, Ch. 7; see also NRL-6598, November 21, 1967.
HEAVY-SECTION STEEL TECHNOLOGY PROGRAM: INTERMEDIATE-SCALE PRESSURE VESSEL TESTS * R.H. BRYAN, J.G. MERKLE, G.C. SMITH and G.D. WHITMAN Oak Ridge National Laboratory, Oak Ridge, Tennessee 3 7830, U.S.A. Received 22 September 1977 Eight flawed 990-mm-OD, 152-mm-thick vessels of ASTM A508 class 2 forging steel or A533, grade B, class 1 steel plate have been pressurized hydraulically to burst or rupture. The rupture test of one vessel (V-7) was repeated pneumatically to study the effects of the sustained load thus attainable. Test temperatures ranged from 0 to 91°C so that transitional to upper shelf fracture toughness behavior was observed. Pretest predictions of failure pressures were based on tensile and toughness properties determined from Charpy-size to 102-ram-thick specimens representative of the test vessels. Linear elastic fracture mechanics based on strain and plastic instability analyses were generally adequate for determining the failure pressures, which ranged from 2.15 to 3.28 times design pressure.
ceived to provide conditions o f size and state o f stress similar to those that would obtain in real reactor vessels in normal operation and under accidental conditions. Although ITVs are intermediate in size between laboratory specimens and reactor vessels, they are nearly full size with respect to thickness and are capable o f providing conditions of transverse restraint not attainable by other means. The ITV tests were planned to provide a connection between the fracture behavior observed in smallscale tests and the behavior o f full-scale components. When the first two ITVs were tested it was not known whether and within what limits linear elastic fracture mechanics (LEFM) would be a useful analytical method. The general plan for the series was to observe flaw behavior quantitatively under a variety o f fracture toughness conditions ranging from low transitional to upper shelf toughness. Some consideration in the early order o f tests was given to demonstrating qualitatively that a margin o f safety in flawed vessels does exist. All o f the test conditions were selected to be related to some normal or hypothetical abnormal phase o f nuclear power plant operation. However, experimental considerations generally dictated the
1. Introduction Development and validation o f methods o f analyzing full-scale light-watercooled reactor pressure vessels containing flaws is a major objective o f the heavy-section stell technology program. Such methods are an essential tool for the evaluation o f the safety o f nuclear power plants. It is especially desirable that estimates o f the strength o f flawed structures be made realistically and quantitatively to ensure adequate margins o f safety in the interest o f public safety and to avoid unnecessary conservatism in the interest of economy and productivity. The testing o f intentionally flawed steel vessels is a necessary part o f this development effort. The interme. diate-test-vessel (ITV) series o f experiments was con* Work done at Oak Ridge National Laboratory, operated by Union Carbide Corporation for the Energy Research and Development Administration. This work funded by U.S. Nuclear Regulatory Commission under Interagency Agreements 40-551-75 and 40-552-75. Expanded version of Invited Paper G1/I* presented at the 4th International Conference on Structural Mechanics in Reactor Technology, San Francisco, California, 15-19 August 1977. 47
R.H. Bryan et aL / Intermediate-Scale pressure vessel tests
48
flaw size; and most tests were carried to the ultimate capacity of the testing system or to the point of structural failure. Extremes of temperature and pressure not directly related to any reactor operating condition were used as necessary to obtain experimental results by which methods of analysis could be decisively evaluated. The tests performed to data have been primarily concerned with the study of the onset of flaw instability [ I - 5 ] . Any method of analysis that is successful in predicting the threshold of instability must be able to relate the geometry of the flaw, structure, and loading to measurable material properties. This usually involves consideration of more than a single fracture criterion. To the extent practicable all of the variables that might have a bearing on flaw behavior were measured during each test: strains near and far from the flaw, crack-opening displacement (COD), crack depth, temperature, and pressure. In addition to the quantitative results of the tests some important conclusions regarding modes of behavior were reached. Conditions for stable crack growth, mixed mode crack propagation, and leak without break were observed.
2. Description of intermediate test vessels Ten 152-mm-thick vessels were fabricated to the configurations shown in fig. l. The 990-mm-OD by 1370-mm-long test sections were fabricated of ASTM A508, class 2 forgings in vessels V-1 through V-6 and of ASTM A533, grade B, class 1 plate in vessels V-7 through V-10. The nozzles in V-5, -9, and -10 were fabricated of ASTM A508, class 2 forging steel. These characteristics are summarized in table 1. The test cylinders of forged vessels V-3, -4, and -6 were cut longitudinally and rejoined by submerged-arc welding to provide weld metal sites for emplacement of flaws as shown in fig. 2. The test section of vessel V-6 also contained an intersecting circumferential weld. The fabrication of the vessels has been described in detail by Childress [6,7]. The vessels were designed and fabricated in accordance with section III of the 1968 edition of the ASME boiler and pressure vessel code. The thickness of the test section was chosen as the minimum adequate for the performance of valid tests of linear elastic fracture mechanics. The length was such that a central flawed region would be unaffected by the dis-
. 380 1D
I
I
I
f
r I- 99o og |
I ~ J'~--152SHELL /II i 1I Jt 123 I "--~'\TEST13170 I J~23~i 10./~ J L-
VESSELS1, 2, 3, 4, 6, AND 7
1
DIMENSIONSIN m
VESSELS5 AND 9
Fig. 1. Intermediate test vessel typical cross sections.
52WALL
49
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
Table 1 Materials used for fabricating intermediate test vessel cylinders and nozzles. Vessel designation
Material Cylindrical shell
V-1 V-2 V-3 V-4 V-5 V-6 V-7 V-8 V-9 V-10
A508 class 2 A508 class 2 A508 class 2 A508 class 2 A508 class 2 A508 class 2 A533-B class 1 A533-B class 1 A533-B class 1 A533-B class 1
Nozzle
A508 class 2
A508 class 2 A508 class 2
continuities at the two heads. The diameter of the test section and closure forging were a practical minimum allowing personnel access for inspection, flaw preparation, and instrumentation. The ITV head and closure are designed in accordance with section III of the code for a design pressure of 138 MPa. In addition, the allowable stress intensity for secondary
membrane plus bending in these components was arbitrarily set at 1.1. times the stress intensity that would obtain in an infinite cylinder at a pressure o f 138 MPa. These measures were taken to ensure that initial failure would take place in the test section. Design parameters of the test vessels are compared with those of typical boiling-water reactor (BWR) and pressurized-water reactor (PWR) pressure vessels in table 2. The ITV thickness is almost full scale with respect to real reactor vessels. Although the ITV design pressure of 66.9 MPa is markedly different from that of the reactor vessels the stress states at design pressure are similar; by virtue of the code definition of allowable stress intensity, the difference between the largest and smallest principal stresses is the same in the cylindrical sections of both types of vessels under pressure loading.
3. Test preparations Intermediate vessel tests were preceded by intensive preparations to obtain a quantitative basis for predicting failure conditions and for evaluating meth-
9e° ee 00~
99° ~'~"O0~ • ONGI,UD,NAL
\
I LONGITOOlNAL SUBMERGED-
~
,.c
1370 mm
685, mm[
J
1370 mm
iii LOCATION
GIRTH SUBMERGEDARC WELD
LOCATION LONGITUDINAL SUBMERGEDARC WELD J
(a)
(b)
Fig. 2. Weld orientations for flaw location. (a) vessels 3 and 4; (b) vessel 6.
50
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
Table 2 Nominal design parameters of cylindrical shells of intermediate test vessels and BWR and PWR vessels. Vessel
Inside Diameter (ram)
ITV BWR PWR
Thickness (mm)
686 6100 4370
Design pressure a fMPa)
152 164 216
Average stresses at design pressure fMPa)
66.9 9.64 17.3
Circumferential
Axial
151 179 175
61.7 87.3
83.6
a Code allowable for a cylinder as determined by the general primary membrane stress intensity.
ods o f fracture analysis. Material properties of the test vessels were measured, and flawed epoxy and steel models were tested. Tentative test conditions were evaluated by experimental and mathematical analysis to determine the specific test conditions, in terms of flaw size and vessel temperature, likely to produce useful intermediate vessel test results.
O 540 A I "1 20 ~
V 1 V-2 V-?
A
V
/,,
f
,,
- Y~- . . . . . . . .
I
Tensile, impact, and fracture toughness properties of the ITVs were determined prior to testing. Prolongations of the test cylinders and nozzle forgings were acquired for this purpose. Thus the material properties were determined for pieces that had experienced the same fabrication history as the vessel test sections up to the time of welding of vessel components together. Prolongations of the forged cylinders with longitudinal welds (V-3, V-4, and V-6) were separated from the vessel test sections after those welds were stress relieved. The accuracy of structural and fracture analysis is
TENSILE TEST TEMPERATURE ROOM ROOM 933°C
~
5~'
4. Material characterization
~
/
---
/o
600
500
o
'
o
/
O
¢
----
©
f
J
I
J
~
°-'
/
f
~460
4oo Lu
(
/
~ 300
~7 V
200
V
100
400
50 100 DISTANCEFROMOUTSIDESURFACEOF CYLINDER(ram)
150
Fig. 3. Lower yield stress as a function of thickness from prolongations of vessels V-l, V-2, and V-7.
0
0
2
4 6 TRUE STRAIN I%)
8
10
Fig. 4. Typical uniaxial stress-strain curve from prolongation of vessel V-1.
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
51
Table 3 Selected values of Klc d used in pre-test analysis. Vessel No.
Flawed material Test Temp. ¢C)
1 2 3 4
54 0 54 24
5 6
88 88
7 7A 7B
91 88 87
9
24
A508 A508 SA weld SA weld A508 A508 SA weld A508 A533 A533 Repair weld HAZ A508
Range of Values (MN/m3/2) near test temperature
Pre-test analysis value adjusted for section thickness (MN/m3/2)
PCCV
CT a
163-190 157-204 182-202 111-202 181-220 181-255 163-190 192-235 162-220 162-220 220 307 184-223
222-335 182-230 198-315 147-214 287-314 165-264 202-353 229-377 245-447 245-447
342 202 357 220 b 302 265 405 429 331 331
159-298
159-298
a From Mager [11]. b Post-test value 176. limited by the quality of available material property data. The variation of properties with depth were measured where practible. Yield stress data, which are especially important in tests that go well beyond yield, are shown in fig. 3 as a function of thickness for three cylindrical test sections. Except for the case of V-2, which was tested at 0°C, the tensile properties were determined near the respective vessel test temperatures. Fig. 4 shows a typical true stress-true strain relationship for V-1 material. The perfectly plastic and strain-hardening characteristics shown in this figure were important factors in analyzing the vessels. Charpy impact energy curves were determined as a function of depth and orientation. Results were typical of A508, class 2, and A533, grade B, class 1 materials. NDT temperatures determined by dropweight tests of shell course material were typically - 1 2 ° C and - 5 1 °C, respectively, for the two materials. Fracture toughness was estimated from test data from bend tests o f precracked Charpy-V (PCCV) specimens and from tests of 21.6- and 76- or 102mm-thick compact tension (CT) specimens. The 76and 102-mm CT specimens were the largest that could be cut from nozzle and shell prolongations.
Data were converted to lower bound Kicd values by the equivalent energy procedure proposed by Witt and Mager [8]. Such calculations have been shown by Merkle and Corten [9] to agree with J-integral calculations, for the same chosen measurement point, for these two specimens. The thickness of the smallest CT specimens corresponds to the thickness of 1/7scale steel model vessels used in this program. Variations of toughness with depth in cylinder wall and temperature were determined for the material in which the flaw would be located. Variability of KIcd values was substantial, as had been expected from earlier studies by Mager [ 10]. Selected properties used in pretest analyses are shown in table 3.
5. Model testing Scale models were used extensively as an aid to planning and interpreting results of tests of the ITVs. Epoxy models, which behave elastically, as well as steel models, which exhibit elastic-plastic behavior, were tested. In each case the model contained machined notches that were sharpened either by fatigue (in epoxy) or by hydrogen-charge cracking of an electron-beam weld placed at the root of the
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
52
/
CIRCULAR
\A " ~AT'~UEOW \~
//"-"~\M.ACHINED ~
IV~
I / / / / / / /,1 ?-\~Af"!~EOA l/ / / / / / / /'L/FA'TI(SUED/ / I 1//// / / I1//// //A
/',~,
\ "~\\\\\1
VESSELS 1, 2, 3, 4, AND 6 I~
2b VESSELS 5 AND 9 _
'Y//]
°.c.,..°
c..c.,° 111
l
I
I
I
I
t'/////h
~Zb °
VESSELS 7, 7A, AND 7B
Fig. 5. Flaw geometries for intermediate test vessels.
Table 4 Flaw design for intermediate-size vessels Vessel
Flaw
Location a
Material
Dimensions (mm) Estimated a
Ab B
Outside Outside Outside Outside Outside Inside nozzle
A508-2 A508-2 SA Weld SA Weld A508-2 A508-2
Actual 2b
a
2b
64 64 65 66 66 30
206 206 210 210 204
65 64 54 75 79
210 211 216 210 204
47 44 45 135 135 140 30
129 132 135 472 472 472
47 34 49
133 132 135
139
467
corner
Ab B
C 7 7A 7B 9
Outside Outside Inside Outside Outside Outside Inside nozzle corner
SA Weld A508-2 SA Weld A533-B1 A533-B1 Sect. XlWeld HAZ A508-2
a Plane of flaw lies in radial-axial plane of test cylinder (and also of nozzle in V-5 and V-9). b Flaw at which fracture occurred.
53
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
machined notch (in steel). The models were instrumented to provide pressure-strain data up to the time of failure. These data provided a means of qualifying various computational methods for determining the strength of flawed vessels as a function of flaw geometry and toughness. Shape factors needed for application of LEFM were measured in epoxy model tests.
6. Test vessel preparation Sharp flaws of the desired shape and location were necessary to satisfy the assumptions of LEFM and to maximize the chances of successful correlation of experiment and analysis. Flaw geometries for the tests are defined by fig. 5; and the estimated and actual flaw dimensions are given in table 4. In all cases the actual initial flaw geometry was close enough to the estimated value to meet test objectives; however, the discrepancies were in some instances large enough to affect the strength of the vessel and accuracy of pretest analysis. Flaw preparation involves two steps: (1) a notch is machined at the desired location to within a few millimeters of the desired final crack tip contour; (2) the machined notch is extended either by fatigue or by a combination of electron-beam (EB) welding and hydrogen-charge cracking. Fatiguing the crack by cyclic pressurization of the notch surface was accomplished satisfactorily for all ITV flaws except those of the VV-7 type. In the V-7 vessel the volume of the machined notch was so large that it was impractical to pressurize it cyclically; furthermore, uniform fatigue crack growth could not be assured around the periphery of a long deep flaw. Therefore, in V-7, V-7A, and V-7B, the machined notch was sharpened by EB welding the periphery to a uniform depth and subsequently hydrogen-charging the brittle weld until a sharp crack was formed. Flaw sharpening in each case extended the machined notch generally by 8 to 25 mm to a smooth final crack contour. Vessel V-7 has been tested three times. This was considered feasible since the test section did not yield through the thickness in regions remote from the flaw. Preparation for the last two tests of V-7, designated 7A and 7B, involved repair of the previous flaw and rupture zone. In each case the cracked region was cut out and the vessel was repaired with-
out post-weld stress relief by the half-bead welding procedure prescribed in section XI of the ASME boiler and pressure vessel code [ 12]. Measurements and studies by Smith [ 13] generally indicate that residual stresses resulting from the section XI welding procedure are low in the weld metal and approach yield in the base metal near the weld. The vessels were extensively instrumented with internal and external strain gages (typically 60 or more), crack-opening-displacement (COD) gages, ultrasonic transducers for monitoring the location of the crack front, acoustic emission sensors, and thermocouples.
7. Test description The intermediate vessel tests were performed under the conditions indicated in table 3. Prior to pressurization, the test vessel was brought to the prescribed temperature range. With one exception, the vessels were pressurized hydraulically until a leak or burst developed. Usually pressure was raised in steps of 3 to 30 MPa, depending on the pressure, with pauses in pumping to allow for recording and assessment of data. Vessel V-7A was pressurized pneumatically in an attempt to replicate the hydraulic test of V-7. The pressurization schedules for the hydraulic tests were generally short, except for the first test, in which trouble with equipment was encountered. Fig.
3o351
240 220 2OO 180 ~. 160 v
o2o I
/I.
i
. . . . . . . . . . .
15
i
0
1
! l
2
=
120
~
100 i
io ? .................
140
3
--
80
-6o
0
Time ( h r )
Fig. 6. Pressure-time record for hydraulic pressurization of ITV-3.
R.H. Bryan et aL / Intermediate-Scale pressure vessel tests
54 200
. . . .
r
. . . .
I
. . . .
I
. . . .
I
. . . .
I
. . . .
I
.
.
.
.
.
.
.
.
I
. . . .
I
. . . .
150
PE
30
IO0
5O
/
/
/1
0
t -10
0
. . . .
d 10
. . . .
J 20
,
,
,
,
I
,
,
,
,
30
l
L
,
40 TIME
,
,
I 50
,
,
,
i
,
60
,
,
,
,
I 70
I J i i J 80
90
(hr)
PRESSURIZATION SCHEDULE FOR V - 7 A TEST
Fig. 7. Pressure-time record for pneumatic pressurization of ITV-7A.
6 shows a typical record of a hydraulic test. The pneumatic test schedule was deliberately protracted because of a combination of test requirements and test facility limitations. Several loading-unloading cycles were desired and four were accomplished as shown in fig. 7.
8. Discussion of results The intermediate vessel tests demonstrated that, large flaws notwithstanding, all of the vessels were capable of yielding in the unflawed region of the test cylinder before the onset of fracture. A long, homogeneous cylinder of the diameter and thickness of the test vessels would behave as shown in fig. 8, which is a calculated pressure-strain curve based upon a trilinear approximation to the stress-strain relationship given in fig. 4 with a yield stress of 520 MPa. One can see in fig. 8 that initial yielding occurs, on the inside surface, when p = 0.26 Oy. The cylinder first yields on the outside surface at the gross yield point, where p = 0.37 Oy. The test results shown in fig. 9 for the cylindrical test sections and in fig. 10 for the nozzles illustrate the significance of the strength demonstration. Initial
yielding on the inside surface corresponds to 0.1% outside surface strain on these pressure-strain plots, a condition all vessels attained. Initial yielding occurs when the pressure reaches 2 X design pressure, the latter being 66.9 MPa. Every flawed vessel that burst sustained a pressure of at least 2.7 X design pressure without fracture, and outside surface strains were 0.168% or greater. In the tests of vessels V-5, V-7, and V-7A the vessels leaked but did not burst. Vessel V-7B did not burst, but the preliminary indications from this most recent test are that the flaw was tearing unstably at the maximum test pressure. One of the specific purposes of testing thick vessels was to study the applicability of methods of fracture analysis to the prediction of failure pressure. The limits of validity of linear elastic fracture mechanics (LEFM) was of particular interest. The three types of flaws shown in fig. 5 present three types of problems. The cylindrical vessels with part-circular flaws provided the simplest geometry for analysis and also provide conditions approaching plane strain at the deepest part of the flaw. Tests of flaws of this type were performed first. Nozzle corner flaws are in a region in which both stress concentrations and transverse restraint complicate the analysis. The deep, long-flaw geometry of vessels V-7, -7A, and -7B,
55
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
220
180
I |
/~I
Fully strain hardened-~
which was originally conceived for a specific study o f leak-without-burst, also presents a different type o f problem. The fracture analysis and test results are summarized in table 5, which compares LEFM analyses with test results. LEFM based on strain was generally used by Merkle [2] in the analyses o f the vessels. In this method fracture mechanics is used to compute a failure pressure and strain, Xf, strictly on the basis of elastic behavior. Since this failure pressure is unrealistically high for strains above yield, the estimated real failure pressure is determined as that pressure at which the computed failure strain, Xf, would obtain in an unflawed cylinder. A calculated or measured pressure-strain curve for an unflawed cylinder, as shown in fig. 8, was used to obtain the failure pressure. The results o f this procedure are in good agreement with the two tests, V-2 and V-4, in the transition range o f toughness. These two vessels fractured extensively prior to gross section yielding. Vessels V-1 and V-3 were tested at a temperature (54°C) for which the static fracture toughness was
%,-Onset of strain . . . . . k-Gross yield hardening
160
140
j f - - I ni tia] yielding
~ 12o . ~
.
.
--
i
.
--
~ ! , 1
10o ~-
'
t
I
80
'
6o
1
I
"
"
"
-. . . . . . .]-
4O 20 - -
]
G 0
0.2
0.4
0.6
0.8
1.0
1.2
Outside s u r f a c e c i r c u m f e r e n t i a l s t r a i n
1.4 (%)
Fig. 8. Typical pressure-outside circumferential strain curve for a long unflawed cylinder with intermediate test vessel dimensions.
240
220
200
~~__~-"-'--'-'~ /~.-V-1
~180
uJ 160 O: o.
140
f
V-7, V-7A
120
100
0
0.5
1.0 STRAIN (%)
1.5
2.0
Fig. 9. Outside circumferential strain remote from flaws. Cylindrical intermediate test vessels 1, 2, 3, 6, 7, and 7A. Vessel V-4 responded approximately as V-2.
56
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests 200
I
I
I
180
160
140
120
100
80
~
•
GAGE 89. V S
--
/k GAGE 9Y, V - 9
X 40
L
20
-
o 0
I
I
I
05
,0
,s
zo
STRAIN 1%1
Fig. 10. Outside ckcumferentia] strain remote from flaw. Intermediate test vesselsV-5 and V-9 with nozzles.
known to be high (above 300 MN/m 3/2) but the crack arrest toughness low. Hence crack arrest was not expected. The cracks propagated in a mixed mode fracture beyond the ends of the test sections. Substantial stable crack growth prior to fracture was expected and did occur. LEFM calculations of the critical pressure and strain, taking stable crack growth into account, were quite conservative. The gross section strains at fracture were well beyond yield. Fracture appeared to be related to the attainment of local plastic instability, which was, in turn, dependent upon the extent of stable crack growth. The latter phenomenon must be accounted for quantitatively in an analysis if the method is to be useful for predicting fracture strength after through-the-thickness yielding OCCURS.
Following the test of vessel V-I, the fracture surfaces were examined by scanning electron microscopy, under the direction of Canonico, to determine the fracture mode. Fig. 11 shows the fractured region of the vessel immediately after the test. Figs.
12 and 13 show fracture surfaces removed from the vessel and the essential results of the fractographic study. The region of stable crack growth extends from the sharpened flaw through the midsbction of the wall for some undetermined distance within the dimple-fracture region shown in fig. 12. Beyond the dimpled region, which was about 500 mm long, the central section exhibited cleavage, as shown in fig. 13. The test of vessel V-6 at 88°C gives, further confirmation to the conclusions derived from the V-1 and V-3 tests with respect to crack initiation. At the higher temperature, however, the V-6 fracture was a full shear terminating within the test section. The decreasing load during fracture probably limited the length of the tear. These five cylindrical vessel tests provided an opportunity to compare fracture behavior in two different materials as well as in different toughness regimes. The flaws that failed in V-3, V-4, and V-6 were located in submerged arc weld metal, while the V-1 and V-2 flaws were in A508 tbrgings. The flaws in weld metal behaved essentially the same as those in base metal, but prediction of failure was affected by the greater apparent variability in toughness of weld metal relative to base metal. The preparations for testing the vessels with nozzles, V-5 and V-9, included an elastic three-dimensional finite-element analysis by Krishnamurthy [14] and experimental determination in epoxy model tests by Derby [1,4] of stress concentrations at the nozzle corner and shape factors for nozzle corner flaws. The elastic stress-concentration factors obtained from the finite-element analysis and the epoxy models (2.98 and 2.79 respectively) were substantially lower than that measured in the intermediate vessel tests (~4). The preliminary shape factor deternfinations were deduced from burst tests of flawed epoxy models, for which material Kic was independently measured [4]. Recent frozen stress photo-elastic experiments by Smith et al. [ 15] on identical models agree with Derby's results. As indicated in table 5, the test pressure of the vessels with nozzles, V-5 and V-9, exceeded the fracture pressures predicted by LEFM based on strain by a considerable margin. Although the stress is high at the nozzle corner on account of stress concentration, the transverse restraint on the flaw is low. Nozzle corner circumferential strains reached about 6.5 and
57
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
Table 5 Calculated and test results Vessel No.
Calculated results
Test results
Local plastic LEFM instability pressure (MPa) Failure pressure (/VlPa)
Mode of failure
2 4 9
189 181 93-127
1 3
206 c 217 c
5 6 7 7A
232 c 132 132
7B
132
190 190 132 190 143 d 143 d 183 e 143 d 183 e
Maximum Maximum pressure (MPa) strain (%) a
Load factor b
Failure strain (%)
Transition range thoughness 0.206 Flat 192 0.163 Mixed 183 0.070-0.096 Mixed 185 Static upper shelf toughness 0.345 Mixed 199 0.398 Mixed 214 Static and dynamic upper shelf toughness 0.100 Leak 183 0.479 Shear 220 0.109 Leak 147 0.109 Leak 144 0.109
Leak with continued slow tearing
152
0.19 0.168 1.05
2.87 2.73 2.77
0.92 1.45
2.96 3.19
0.25 2.0 0.12 0.110
2.74 3.28 2.20 2.16
0.115
2.26
a Outside circumferential strain remote from flaw. b Ratio of failure to design pressure. c Assuming 15% stable crack growth. d Ligament rupture pressure. e Burst pressure.
8.5% in V-5 and V-9, respectively, in comparison with predicted strains of 0.35 and 0,6 to 1.1%. Although pretest analyses underestimated fracture pressures and strains in the vessels with nozzles the test results were predicted qualitatively. Vessel V-5, which was tested at 88°C, did not fracture; the crack grew stably with increasing pressure until a leak developed. Fracture pressure is presumably higher than the failure pressure observed in V-5. Vessel V-9 fractured at about the pressure at which V-5 leaked. Because of the low V-9 test temperature (24°C), the arrest toughness was below the static thoughness, and the crack propagated rapidly. The crack had, however, grown stably from a depth of 30 mm to about 50 mm before becoming unstable. Work on nozzle corner cracks is continuing and one additional ITV with a nozzle will be tested. The test of an intermediate vessel with a long,
deep flaw (fig. 5) was initially conceived as a demonstration of leak-without-burst. Accordingly, the size of the flaw in vessel V-7 was chosen so as to attain failure between design pressure (66.9 MPa) and gross yield pressure. Tests of flawed steel models confirmed the latter to be about 190 MPa. Pretest analyses of the two possible modes of failure were made. The leak-without-burst mode of failure obtains if the ligament at the deepest part of the flaw ruptures prior to obtaining a condition at which the flaw is unstable with respect to axial propagation. The predicted conditions for leak shown in table 5 were based upon LEFM with a ligament-size-dependent fracture toughness. This analysis also agreed with model test results. Burst estimates were based upon a LEFM analysis of a through-the-wall crack with corrections made for plastic zone size and bulging. Without considering the
c~
~J
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
ITV Vl ASTM A508 CL 2
~ i!~i~iii~i
Fig. 12. Vessel V-1 fracture surface in region of prepared flaw. The macro photograph shows the machined notch surface, the fatigued crack extension, and the region of ductile fracture. The scanning electron microscope photomicrographs show dimple failure.
59
60
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
ITV V1 ASTM A508 CL 2
Fig. 13. Vessel V-1 fracture surface in region of prepared flaw. The inset shows a region near the termination of the ductile zone which was examined by scanning electron microscopy. The photomicrographs show the transition from dimple to cleavage mode of failure.
elevation o f fracture toughness from lack of transverse restraint the analysis predicted a burst. With an adjustment for less than full transverse restraint according to an empirical relationship o f Irwin [ 16] the analysis indicated that the vessel would attain gross yield prior to burst. It was also estimated that a local plastic instability around the flaw would
develop prior to ligament rupture or burst. The V-7 test was carried out hydraulically with the expected result, a leak at 147.2 MPa. Upon depressurization the vessel resealed itself and held a pressure o f about 125 MPa. The vessel was repaired and retested pneumatically with an identical flaw in a virgin region o f the vessel. A thin-membrane patch was welded to
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
61
Fig. 14. Fracture surfaces from vessel V-7A. The light areas at the sloping ends are surfaces fractured in liquid nitrogen after the test. the inside surface beneath the flaw so as to allow the pressure to be maintained long after rupture. The sustained load test, designated V-7A, produced flaw behavior practically identical to that o f the first test. Rupture occurred at 144.2 MPa, the crack had grown about 10 mm toward each end but did not propagate. The fracture surfaces in V-7A are shown in fig. 14. The light areas at the sloping ends of the flaw were fractured in liquid nitrogen after the test. Fractographic examination shows the areas of crack-extension to be dimple fracture. In both of these tests the location of the edge of the crack was observed ultrasonically. The crack grew stably with increasing pressure, as shown in fig. 15 for the V-7 test. Stability during pressurization was confirmed by observing crack-opening displacement (COD), o f which fig. 16 shows typical behavior and a comparison of V-7 and V-7A. This indicates that COD continues to increase to some degree during periods of constant or slightly decreasing pressure, a phenomenon that was observed in all of the intermediate vessel tests. This creeping
(in.)
(rnm}
1.2
1.0
0.8
0.6
0.4
0.2
0 (MPa)
34,5
~.9
1Q~.4
I
I
I
I
15,1000
20,000
,6000
10,000 PRESSURE
137,9
Fig. 15. Crack depth and inside surface deformation versus pressure measured ultrasonically in vessel V-7.
I
R.H. Bryan et al. /Intermediate-Scale pressure vessel tests
62 (mrn)
(in) 06
15
} 0.5
uJ 10
I
V-7
V-7A GAGEZT 129 CYCLE 2 • CYCLE3 CYCLE4
0.4
v) C3
03
5 c,
~s 5
0.2
/
!
•
t
J
01
(psi)
(MPal
0 6000
I
10,000
I
50
14,000
18,000
I
22,000
Ti
100 PRESSURE
150
Fig. 16. Crack-opening displacement versus pressure for vessels V-7 and V-TA,
behavior, which had been suspected as possibly indicating incipient instability, was carefully evaluated in the test of V-7A, which was pressurized pneumatically and specially instrumented for this purpose. The stability of the crack at peak pressure is indicated by the asymptotic behavior of the COD at maximum pressure shown in fig. 17. The maximum pressure was maintained within 1.1% after rupture occurred. The third test of this vessel, called V-7B after the second weld repair, was performed by hydraulic pressurization. The flaw geometry was identical to that in the V-7 and V-7A tests. The sharpened flaw in V-7B was placed as nearly as possible within the heataffected zone resulting from the repair. A pressureretaining patch was also installed in this vessel so that the sustained-load behavior o f the flaw in the neighb orhood of the weld repair could be compared with the behavior observed in the previous tests. The vessel ruptured at 152 MPa with outside surface strains remote from the flaw very nearly the same as in V-7 and V-7A. The patch allowed the pressure to be maintained almost constant for more than
a minute after rupture was observed. At the time of rupture and during the period of sustained loading thereafter, the COD increased much more rapidly than at lower pressures and continued to do so until the patch ruptured and the vessel depressurized. Present interpretations of the behavior of V-7B are 4.6
,..,.,..r,-,
...r ,.¸ , . - .
-,-..,
......... r'--'"',"-,--,..-.,..-,
-.,..-,--.
-,....,--,-..,
....
4.4 4.2
f
v~4"O~
/
~3.8
/
3.6
~ T l m e ' of rupture
/ 3.4 / 3.2
i
400
800 1200 Time( sec )
i600
2000
Fig. 17. Crack-opening displacement versus time for vessel V-7A during final period of sustained load.
400
R.H. Bryan et al. / Intermediate-Scale pressure vessel tests
based upon preliminary evaluations of the data and incomplete examination of the flaw. In V-7B a short tear through the ligament is visible, whereas previously the rupture could not be seen until the flaw was cut from the vessel. In V-7B crack extension axially is visible for at least 50 mm at both ends, and there are preliminary ultrasonic indications that the crack runs out o f the original plane o f the flaw. The data suggest that the vessel was slowly tearing just prior to depressurization. It has not yei been determined how far the crack had grown before the COD became unstable. However, the pretest burst analysis shows that a through-the-wall crack ! 80 mm longer than the initial flaw would be unstable at 152 MPa.
9. Summary and conclusions The tests of intermediate-size vessels with sharp flaws permitted the comparison of experimentally observed behavior with analytical predictions o f the behavior o f flawed pressure vessels. Fracture strains estimated by LEFM were accurate in the cases in which the flaws resided in regions of high transverse restraint and the fracture toughness was sufficiently low for unstable fracture to occur prior to yielding through the vessel wall (viz V-2 and V-4). When both of these conditions were not present, unstable fracture did occur, always preceded by stable crack growth; and the cylinders with flaws initially less than halfway through the wall attained gross yield prior to burst (viz V-l, V-3, and V-6). Predictions of failure pressure of the vessels with flawed nozzles, based upon LEFM estimates of failure strain, were very conservative. LEFM calculations of critical load were based upon small-specimen fracture toughness test data. Whenever gross yielding preceded failure, the actual strains achieved were considerably greater than the estimated strains at failure based on LEFM (viz V-l, V.3, V-6, and V-9). In such cases the strength o f the vessel may be no longer dependent upon plane-strain fracture toughness but upon the capacity of the cracked section to carry the imposed load stably in the plastic range. Stable crack growth, which has not been predictable quantitatively, is an important factor in elastic-plastic analysis of strength. The ability of the flawed vessels to attain gross yield in unflawed sections has important qualitative
63
implications on pressure vessel safety margins. The gross yield condition occurs in light-water-reactor pressure vessels at about 2 × design pressure. This loading condition is generally well beyond the allowable normal, upset, and emergency conditions for which reactor pressure vessels may be designed. The intermediate vessel tests that demonstrated a capacity for exceeding this load (vis all except V-2, V-4, and V-7) confirm that the presumed margin of safety is not diminished by the presence of flaws of substantial size, provided that material properties are adequate. References [ 1] R.W. Derby et al., USAEC Report ORNL-4895, Oak Ridge National Laboratory, February 1974. [2] R.H. Bryan et al., USAEC Report ORNL-5059, Oak Ridge National Laboratory, November 1975. [3] J.G. Merkle et al., ERDA Report ORNL/NUREG-1, Oak Ridge National Laboratory, August 1976. [4] J.G. Merkle et al., ERDA Report ORNL/NUREG-7, Oak Ridge National Laboratory, August 1977. [5 ] R.H. Bryan et al., ERDA Report ORNL/NUREG-9, Oak Ridge National Laboratory, to be published. [6 ] C.E. Childress, USAEC Report ORNL-TM-4351, Oak Ridge National Laboratory, October 1973. [7] C.E. Childress, Documentary Report 5, USAEC Report ORNL-TM-5074, Oak Ridge National Laboratory, January 1976. [8] F.J. Witt, T.R. Mager, USAEC Report ORNL-TM-3894, Oak Ridge National Laboratory, October 1972. [9] J.G. Merkle, H.T. Corten, J. Pressure Vessel Tech. ASME, 96 J. (1974) 286. [10] T.R. Mager, Report No. WCAP-7578, Westinghouse Electric Corporation, PWR Systems Division, Pittsburgh, Pennsylvania, October 1970. [11] T.R. Mager, S.E. Yanichko, L.R. Singer, Report No. WCAP-8456, Westinghouse Electric Corporation, Nuclear Energy Systems, Pittsburgh, Pennsylvania, December 1974. [12] "Rules for Inservice Inspection of Nuclear Power Plant Components," Section XI, Subsubarticle IWB-4420, 1974 edition with Summer and Winter 1974 Addenda, American Society of Mechanical Engineers, New York. [13] G.C. Smith, ERDA Report ORNL/NUREG/TM-94, Oak Ridge National Laboratory, pp. 39, April 1977. [14] N. Krishnamurthy, USAEC Report ORNL-TM-3315, Oak Ridge National Laboratory, March 24, 1971. [15] C.W. Smith, M. Jolles, W.H. Peters, VPI-E-76-25 (also ORNL/Sub/7015-1) Virginia Polytechnic Institute and State University, November 1976. [16] G.R. Irwin, et al., USAEC Report ORNL-NSIC-21, Oak Ridge National Laboratory, December 1967, Ch. 7; see also NRL-6598, November 21, 1967.