TANAKA ET AL ON NEW WIDE PLATE ARREST TEST 25- 123docz.net

I Weld / IV /Fatigue crack 1 metal ~L. ~ development

/ ~80~85~ during the

Fatigue CTOD test

a. Side cracked specimen

(

We'd--/. I /

='<"/H /-e I

~ Fatigue CTOID

test 1

I b. Face cracked specimen

FIG. 2 - - F a t i g u e C T O D test on weld joints.

Table 1 lists examples of the monotonic and fatigue CTOD test results for weld HAZs for three kinds of low temperature steels. Numerous monotonic CTOD tests were conducted on each of the weld joints. The results are shown in the form of the coefficients in the Weibull distribution equations; these fit well the test results. For comparison, 6~ values for 2% cumulative probability were taken as the minimum ~ values of the weld joints. The fatigue CTOD test showed results conforming to those for the monotonic CTOD test. This fact indicates the possibility of economization of fracture toughness tests by means of the new test method without losing, or instead with increasing, relia- bility. Fracture initiation points in the fatigue CTOD test were also the grain- coarsened HAZ close to the fusion line. Moreover, it is surprising to find that the low-carbon grain-refined steels showed 6~ values of 0.04 mm or lower when some unfavorable combinations of metallurgical and testing conditions were met, since these steels have showed satisfactory performance in all con- ventional tests and notched wide plate tests as well as in actual application to L P G storage tanks for more than 20 years.

Significance of Local Brittle Zones

In small-scale tests for fracture initiation toughness, such as the CTOD and Kic tests, the first incident which leaves a discontinuity in the load versus displacement record is taken as the critical point for investigation. With these tests, however, it is difficult to clarify the significance of the incident with regard to the total safety of the structures. In order to assess the low ~c values Copyright by ASTM Int'l (all rights reserved); Wed Dec 23 18:28:35 EST 2015

26 F R A C T U R E M E C H A N I C S : S E V E N T E E N T H V O L U M E

[..,

I'N

~ N

~-~ ~

b , I

= _~

~ g

[...,

9 . ~ .

9 o o

I I I

. ~

t ~

j - ,

z ~ + ~

+ -

u'b

I ~_ + ~ ,~ 3.=

"~ . Z .

H o ~ q ~

= o s~w

, , ~-~

TANAKA ET AL ON NEW WIDE PLATE ARREST TEST 27

found in the monotonic or fatigue CTOD test, therefore, surface-cracked wide plate tests were carried out on one of the welded joints that showed low 6r values. Two specimens were prepared from the welded joint described in the fourth line of Table 1. A full breadth surface crack was produced by cyclic bending so that the fatigue crack tip lay in the same brittle zone as in the CTOD specimens. One of the two specimens, when tested at --50~ by a 20 MN capacity tension test rig, showed brittle fracture with the stress lower than the yield stresses of the base and weld metals; the other specimen showed brittle fracture with a high stress. Analysis on the crack tip location of the latter specimen showed that the initial fatigue crack was too long and had its tip in the grain-refined HAZ. The fracture surface of the former specimen and its data are shown in Fig. 3. There was no indication of brittle fracture arrest. Subsequent study revealed that the fracture initiation point of this specimen was the same grain-coarsened region as in the CTOD specimens with the low 6r result. The linear elastic fracture mechanics (LEFM) calculation for this wide plate specimen led to a 6r value of 0.032 mm, which is almost consistent with the values from the monotonic and fatigue CTOD tests.

Agreement of 6r values from CTOD tests and the wide plate test suggests that the results from the small-scale fracture toughness tests can be applied to prediction of the fracture initiation stresses of fatigue-cracked wide plate test specimens that simulate actual structures with fatigue crack defects. How- ever, an infinite surface crack such as that used in the wide plate test is not expected in actual structures. It is important to investigate whether or not brittle fracture can propagate in the unnotched weld HAZs.

Development of the Short Crack Arrest (SCA) Test

The ESSO test has been used widely for the evaluation of brittle fracture arrest capability of steel materials including base plates and welded joints.

The present authors, however, had quite often unexpected results from tests on welded joints in the as-welded condition. Figure 4 illustrates examples of fracture paths of as-welded joints. It is considered at the present time that the compressive weld residual stress, in the direction transverse to the weld line, at the specimen edges must be the cause of this deviation, since it is not observed in welded and center notched wide plate tests. This compressive weld residual stress at the edge is considerably high (Fig. 4). Since this situation is not expected in actual structures, modification of the test method is neces- sary.

The basic idea for the approach taken here is that the brittle fracture has to be initiated at the center of the specimen width to eliminate the effect of the compressive residual stress. The present authors recognize the importance of the concept proposed by an ASTM committee [8] in 1960. In this concept the arrest of a brittle fracture becomes possible, when it propagates just beyond the initial surface crack, due to the higher toughness associated with plane Copyright by ASTM Int'l (all rights reserved); Wed Dec 23 18:28:35 EST 2015

2 8 FRACTURE MECHANICS: SEVENTEENTH VOLUME

t,.,

~t~ o

TANAKA ET AL ON NEW WIDE PLATE ARREST TEST 29

~ G

1,0

Paths.

T e s t ~ weld Loading direction

~ 0 . 4 0 0.60

1 - 2 0 0

RS/ETy

-0.6 0

0 MPa a. F r a c t u r e paths in the

ESSO t e s t s on a welded joint.

b. Weld residual stress distribution along weld line.

FIG. 4--ESSO test results for weld H A Z of a low-carbon fine-grain low-temperature steel

stress situation (Fig. 5a). When this concept is applied to the weld joints in a large panel such as the side shell in an LPG storage tank, one may expect the run-and-arrest phenomenon of brittle fracture (Fig. 5b). Here, possibilities of three types of brittle fracture arrest are pointed out: the pop-in, the SCA, and the long crack arrest (LCA) types. The pop-in type arrest is attributed to the local brittle region at the crack tip and the high arrest toughness of the neigh- boring material. The LCA may be attained because of the high arrest toughness of the base plate. As was indicated by Irwin [ 9], the effect of bulging and possibilities of brittle fracture re-initiation and ductile fracture should also be considered in the LCA concept. In the case of thin-walled vessels, the LCA concept seems not to be practicable.

After these considerations and experimental efforts, a new wide plate test method, the SCA test on weld HAZ (HAZ-SCA test), has been developed. The specimen geometry, test method, and an example of fracture appearance are shown in Fig. 6. The test method applies a brittle starter plate made of 0.45%

carbon steel which initiates a brittle fracture at the notch tip in it and injects the fracture into the specimen. This starter plate has a geometry similar to a compact specimen and is welded by a brittle welding material on a specimen surface with its crack line meeting the weld HAZ of the specimen. Before welding the starter to the specimen, brittle weld beads are planted beneath the starter (Fig. 6). This is made in order to secure the penetration of the brittle fracture into the specimen. A 120 mm long surface notch has been machined with its tip along the weld HAZ for two purposes: (1) to guide the brittle crack so that it runs at least once along the HAZ, and (2) to simulate a surface defect that may exist in actual structures. This surface crack ensures Copyright by ASTM Int'l (all rights reserved); Wed Dec 23 18:28:35 EST 2015

30 FRACTURE MECHANICS: SEVENTEENTH VOLUME

a . Sketch of f r a c t u r e origin (1), slow growth to (2),

fast f r a c t u r e to (3) with a r r e s t due to toughness associated with plane-stress situation, ~ 8

de+ec, S

Z_.l_..Ti c c c ~ c c / c c c c c

| | ~) |

(j] Immediate arrest (pop-in)

(~ Short Crack A r r e s t (SCA) (~ Long Crack A r r e s t (LCA)

Chances of b r i t t l e f r a c t u r e a r r e s t for welded joint FIG. S - - A r r e s t of brittle fracture initiating in a weld joint.

formation of a through-the-thickness crack with the same length as that of the surface crack (Fig. 6c). In the test a brittle fracture is initiated at the notch tip in the starter by loading it with a small jack having the capacity of 100 kN.

The brittle crack travels through the brittle beads and penetrates into the specimen. Test results are divided into " G o " and "Arrest". Unless the brittle crack propagated to a total crack length more than the sum of the surface notch length and twice the plate thickness, the result is judged as "Arrest".

HAZ-SCA Test Results

The new test method has been applied to a wide range of steel materials for low-temperature and cryogenic application. Several specimens were prepared for each series of welded joints, and the test temperatures and the applied stresses were selected so as to obtain a critical temperature versus stress transition curve for each series. Examples of the HAZ-SCA test results and the transition curves are shown in Fig. 7. Fracture appearances of the test data indicated by (a), (b), and (c) in Fig. 7 are displayed in Fig. 8. Photograph (a)

TANAKA ET AL ON NEW WIDE PLATE ARREST TEST 31

FIG. 6 - - T e s t sequence and fracture surface of HAZ-SCA test.

32 FRACTURE MECHANICS: SEVENTEENTH VOLUME

500

400 n

300

,4,.a

(D 200

< 100

Test Base plate Joint result material type Go Arrest

Welding Welding method material ( ~ ) J [ ] Low C-fine grain

II low temp-steel A ~ / , ' / / k 3.5%Ni (A-203F) - 9 (~)s SS O 3.5%Ni (A-203F)

L o w a l l o y e d

X SMAW(3G) (Ti-B) K SMAW(3G) 6Ni X SMAW(3G) 6Ni

. , - -

0 1 k~uu ~ I i I J I i i

- - 8 0 - 6 0 - 4 0 - 2 0

Temperature, ~

FIG. 7--Examples of HAZ-SCA test results.

in Fig. 8 is an example of the test result of " G o " , whereas (b) and (c) are examples of "Arrest".

When the fracture surfaces were investigated carefully, it was clear that, even though the fracture was originated by the starter, the fracture propagation pattern (except the central part of the surface notch) resembled to quite an extent that of the ordinary brittle fracture initiating at the tip of a surface crack. The specimen shown as (b) in Fig. 8 was taken from the same weld joint as that for the full breadth surface crack wide plate test shown in Fig. 3.

From these test results, it became clear that brittle fracture initiating at a tip of a surface crack can be arrested when the length of the surface crack is not so long that the driving force of the through-the-thickness crack exceeds the fracture arrest toughness of the full thickness material which includes the plane stress regions near the plate surfaces.

In Fig. 7 the significant effect of the weld joint type is observed. The K-joint of 31/2% nickel steel shown as No. 2 has a critical temperature 20~ higher than that for the X-joint of the same steel shown as No. 3. The reason for this is that the weld HAZ of the X-joint is inclined against a line perpendicular to

TANAKA ET AL ON NEW WIDE PLATE ARREST TEST 33

a. (Upper). 31/~z~ Ni steel, K-joint, 6Ni-WM, tested at --'/0~ 392MPa.

b. (Middle). LowC-fine grain low temperature steel, X-joint, Ti-B-WM, tested at --50~ 147MPa.

c. (Lower) 3 ~ % N i steel, X-joint, 6Ni-WM, tested at --50~ 392MPa.

FIG. 8--Examples of fracture appearances of HAZ-SCA tests.

34 FRACTURE MECHANICS: SEVENTEENTH VOLUME

the plate surface, whereas brittle fracture tends to propagate in a plane perpendicular to the plate surface. It is also observed that the effect of the nickel content in the base metal is noticeable. This effect was investigated further by testing a series of steel plates having various nickel contents. Figure 9 illustrates the results of this investigation where the same K-type joint shape and the same welding condition were applied for all the materials. It is seen that the brittle fracture propagation along the weld HAZ at an applied stress be- low yield took place at --50~ when the nickel content of the base metal was 2V2% or less.

Compact Crack Arrest (CCA) Test

Since the SCA test requires a large-scale testing machine and a sophisti- cated apparatus for the fracture starter, it is not convenient for wider applica- tions at many laboratories. The CCA test is easier to conduct and requires less material. It is also advantageous that the fracture arrest toughness value, Ka,

is obtained by the CCA test. Because of these conveniences of the CCA test, the present authors also carried out the tests on the same welded joints as for the SCA tests.

i -

U) 4.a

n u) |

L -

U) 500 400 300 200 100 0 500 400 300 200 100 0

a) Fine grain low temp. steel (0.34%Ni) e / o 26turn thick /

,Go / 0 Arrest/

I I I I ( I I I I

- 9 0 - 7 0 - 5 0 - 3 0 - 1 0 Temperature, ~ b) 1.5%Ni steel

25ram thick O /

9 / (~G?rest) U3 4-J

500 400 300 200 100 0 500 400 300 200 100 0

9 9

c) 2~%Ni steel e GO / 30mm thick o Arrest/

I I I I I I I I I

- 9 0 - 7 0 - 5 0 - 3 0 - 1 0 Temperature, ~

J O

eGO / d) 3~%Nisteel OArrest/ 30ram thick

I [ I I I I I I I I I I I I I I I

- 9 0 - 7 0 - 5 0 - 3 0 - 1 0 - 9 0 - 7 0 - 5 0 - 3 0 - 1 0

Temperature, ~ Temperature, ~

FIG. 9-- H A Z - S C A t e s t r e s u l t s f o r w e l d m e n t s o f n i c k e l a l l o y e d s t e e l p l a t e s ( S M A W , 3 G , K - j o i n t , 3 5 to 4 0 k J / c m ) .

TANAKA ET AL ON NEW WIDE PLATE ARREST TEST 35

During the investigation by means of the CCA test the authors made some modifications in the test and the calculation methods. These were as follows:

1. Side Groove--Specimens without side grooves were employed because fracture toughness values of the full thickness were of interest.

2. Fracture Initiation Beads--In order to control the fracture initiation stress intensity factor, KQ, the following welding materials were applied:

9 For the test at a temperature down to --30~ Murex-Hardex.

9 For a temperature between --30 and --100~ Hardfacing electrode for Hv : 300.

9 For a temperature between --100 and --150~ electrode for 800 MPa class strength.

9 For a temperature between --150 and --196~ 6% nickel electrode.

3. Application of Pre-Loading--In order to avoid a premature fracture at a too low KQ value the pre-loading method at a warm temperature was conducted. The same K value as an aimed KQ value was applied at a warm temperature that would not cause fracture initiation.

4. Calculation of Ka Values--Even though the tests were conducted with careful design and preparation of specimens, many specimens showed higher KQ and Ka values which exceeded the limitations set by Ripling et al [10] as a function of the specimen size and the materials' yield strength. In order to utilize as many data as possible, the following modifications in the K , calculation method were made.

9 Calculate the KL value corresponding to the limit load for the specimen; this is obtained by means of the equation by Merkle et al [11].

9 When the experimental KQ value is larger than KL, substitute the following equation for the original Ka formula which uses displacement for calculation of K,. The new formula was obtained by Crosley et al [12] and applies the relationship between initiation K values, crack jump length, and

Ko:

K~ = KL [1 -- 0.92 ( A a / W ) + 0.33 ( A a / W ) 2]

where Aa is the crack jump length and W is the specimen width.

The data obtained in this way were compared with the wide plate test results and were used for the analysis reported in the following section.

Discussion

Effect of Nickel Content of Base Plate on HAZ-SCA Test Results

As was observed in Figs. 7 and 9 the increase of the nickel content of the base metal improves the arrest capability of the weld HAZs. In order to show Copyright by ASTM Int'l (all rights reserved); Wed Dec 23 18:28:35 EST 2015