ORNL/NUREG/TM 171

WSM

Assessment of Lamellar Tearing

J. W. McEnerney

Prepared for the U.S. Nuclear Regulatory Commission

Office of Nuclear Regulatory Research
Under Interagency Agreements ERDA 40-551-75 and ERDA 40-552-75
 ORNL/NUREG/TM-171

Contract No. W-7405-eng-26

METALS AND CERAMICS DIVISION

ASSESSMENT OF LAMiLLAR TEARING

J. W. McEnerney

Manuscript Completed — February 21, 1978

Date Published - March 1978

NOTICE: This document contains information of

preliminary nature. It is subject to revision
or correction and therefore does not represent
a final report.

Prepared for the

U.S. Nuclear Regulatory Commission
Office of Nuclear Regulatory Research
Under Interagency Agreements ERDA 40-551-75 and EROA 40-552-75

Prepared by the 1 nonet

OAK RIDGE NATIONAL LABORATORY I M i NVOVl > • ^IVOJOTtf • M H M M Of Wnt
• • • — i n * * r * t U M M * S u m G m m m i m Wtufcrr * c
Oak Ridge, Tennessee 37830 UlMVf •IWW ttflf W UMQM 9IOW9 DtpSnHOTIt Of
CMffJ^. NOV ttff of Mflf #PJM|BMM» ojoT M f or wm
operated by MMtfKfOfl, OlACOIWM'fltf, Of Mfk CMpfOfin, M H f l
wnf wfmnfy, fioftfi o» MHMJAML or • M U N B pjty bpri
UNION CARBIDE CORPORATION IMtliff of rojpoMftvfy • * nv Kcmcy. ( M f l n f M M
*w iHtfdtaon of my WMtmtMN. M M n t w , ORMVCI OC
for the •TOOfOJ WttlMM, W IffWBW BUt Ht MR WMM MM
Mfr,fl0 pnvojvffy M O M rips.
DEPARTMENT OF ENERGY

m
 CONTENTS

ABSTRACT 1
INTRODUCTION 2
DESCRIPTION 3
MATERIAL VARIABLES 7
WELDING VARIABLES 12
DESIGN VARIABLES 15
TESTS FOR ASSESSING MATERIAL SUSCEPTIBILITY 18
Tests Without Simulated or Actual Welding Cycle 19
Tests With Simulated Teaperature and Strain Cycles 28
Tests With Teaperature and Strain Cycle by Welding 29
Tests of a Prototypic Welded Joint 34
DETECTION AND REPAIR OF LAMELLAR TEARS AFTER WELDING 34
SIGNIFICANCE OF SERVICE LOADING OF MATERIAL SUSCEPTIBLE TO
LAMELLAR TEARING 36
RECOMMENDATIONS FOR AVOIDING LAMELLAR TEARING DURING FABRICATION
OR SERVICE LOADING 41
ASSESSMENT OF ASME CODE REQUIREMENTS 45
CONCLUSIONS 49
ACKNOWLEDGMENTS 50
REFERENCES 50

iii
 ASSESSMEM OF LAMELLAR TEARING

J. V. McEnerney

ABSTRACT

Information on lamellar tearing is summarized and

related to proposed ASME Code requirements.
Lamellar tearing is characterized as a complex phenom­
enon related to poor short transverse ductility and through-
thickness strain. Although nonmetallic inclusions play
a dominant role in lowering the ductility, steel matrix
properties are also important. The through-thickness
strain is generally caused by localized weld thermal
contractions, but could result from certain service loads.
The material, welding, and design variables that affect
lamellar tearing are shown to be complex and interrelated.
The cobnonly reported tests for assessing material
susceptibility are described, with the controversy over
their validity being carefully detailed. Although the use
of a nondestructive test such as ultrasonic examination is
most desirable, a widely applicable test method does not
appear to be available. Of the destructive tests, the
short transverse tensile reduction-of-area currently offers
the most applicable means of assessing material susceptibility.
However, because of the importance of matrix toughness,
the short transverse Charpy V-notch test should be considered
for use as an addiclonal test if acceptance limits are
developed.
The ultrasonic detection of lamellar tears is susceptible
to interpretation errors, which can make it overly conserva­
tive and lead to unnecessary repairs. The repair of tears
is described as costly, difficult, and sometimes ineffective.
Current ASME Code Section 117, Division 1. Subsection NF
design requirements appear to preclude any failures during
static and fatigue service loads. However, without improvement
of short transverse ductility, certain dynamic service loads
could cause lamellar tearing failures. Two alternate
design paths ire recommended to prevent tearing during
fabrication or service loading. The more desirable path
improves short transverse ductility while the other relies
on ultrasonic examination of finished welds or buttering.
The current and proposed ASME Code Section III, Division 1,
Subsection NF requirements dealing with lamellar tearing
are reviewed and recommendations are made.

1
 2

INTRODUCTION

Incidents of laaellar tearing have been reported in most heavy

fabrication industries, such as shipbuilding, nuclear and fossil power
plant erection, bridge and building construction, production of off­
shore structures, and conventional pressure vessel fabrication. These
incidents are normally associated with the welding of hot-roiled carbon
and low-alloy steel plates and shapes aade by ccnventional steelaaking
practices (CON). Table 1 lists soae typical Aaerican and British
1 2
steel specifications with reported ' histories of laaellar tearing.
Because of the widespread occurrence, a considerable aaount of research
has been devoted to investigating the phenomenon. However, despite
this research effort, there is still disagreeaent concerning various
aspects of the laaellar tearing problea.

Table 1. Typical Aaerican and British Steel Specifications

8
with Reported Histories of Laaellar Tearing

Noraal
Specification Description
Deoxidation Practice

British Specifications
B. S. 1501-151 Structural Carbon Steel Seal Killed
B. S. 1501-161 Structural Carbon Steel Fully Si-Killed
B. S. 1501-224 Craln Refl .ed Structural Fully Si-Killed plus Al
Carbon Steel
B. S. 4360-43 Structural Carbon Steel Seal Killed
Lloyds E Grade Crain Refined Structural Fully Si-Killed plus Al
Carbon Steel
Admiralty "B" Grain Refined Structural Fully Si-Killed plus Al
Carbon Steel
Aaerican Specifications
SA-, A 36 Structural Carbon Steel Seal Killed
SA-, A 283 Structural Carbon Steel Si-Killed
SA-, A 285 Pressure Vessel Carbon Seal Killed
Steel
SA-, A 515 Pressure Vessel Carbon Si-Killed
Steel
SA-, A 516 Grain Refined Pressure Fully Si-Killed plus Al
Vessel Carbon Steel

J.CM. Farrar and R. E. Dolby, "Laaellar Tearing in Welded Steel

Fabrication," Phase I of Invejtigatione into Lamellar Tearing. A Compendium
of Reports from a Sponsored Research Prograrne, Welding I n s t i t u t e , Cambridge,
England, March 1975; S. Canesh and R. D. Stout, "Material Variables Affecting
Laael'ar Tearing Susceptibility in S t e e l s , " Weld. J. (Miami) 55(11):
341-s-354-s (Novcaber 1976).
 3

An ASHE Code committee task group has been considering the problem
of lamellar tearing for seveal years. They are currently formulating
requirements for ASHE Code Section III, Division 1, Subsection NF compo­
nent support applications. The intent of this report is to summarize
the available information on lamellar tearing and then examine its
significance and the effectiveness of the proposed ASHE Code III-MF
requirements.

DESCRIPTION

Lamellar tearing is characterized by a steplike cracking, which

occurs parallel to the rolling plane in susceptible steels that have
experienced large strains in the through-thickness direction. Figure 1
1
shows a schematic representation of lamellar tearing. The large
strains usually result from the restraint of weld thermal contractions
and therefore are highly localized. However, the strains could also
result from loading a component in the through-thickness direction
during service.
Lamellar tearing normally occurs in certain welded structures
vhich are highly restrained. The restraint can be imposed by a massive
component or by a smaller one which has been stiffened. Lamellar
! 3 %
tearing is commonly reported ' ' to occur in the following types of
structures: nozzle or penetration through a rigid plate, cylindrical
stiffeners or closure plates, box structures, and stiffened structural
1
joints such as beam-to-column. Figure 2 shows some examples '"* of
these structures. For highly susceptible material, tearing has been
1
reported for low-restraint conditions, such as pullout of lifting
lugs and flange-to-web joints in the fabrication of I-beams.
The two basic joint designs that are subject to lamellar tearing
are T- (cruciform considered more severe form of T) and corner
3,< 1
joints.'» * Figure 3 shows examples of these joint designs. These
joint types provide high restraint because weld thermal contraction
is opposed by compressive forces in the base metal of the components
u
being joined. Although the butt weld is not considered to be particu­
1 3
larly susceptible, some instances of tearing have been reported. '
 4

ORNl-UWG 77-19303

Fig. 1. Schematic Representation of Laaellar Tearing Resulting

Froa Weld Thermal Contraction in a Full Penetration T-Joint.

m u n i oa fMWaTOM cnncM. jomn

>»fic»T«> rmc
WKUBrun

oMcuaMMorm.

CnMIKH vtSMl

tmrKiiamT

CMfTtCM. iOf%T

CMITIOtl JOKTO

Fig. 2. Structures in Which Lamellar Tearing Commonly Occurs, (a)

Penetration through a rigid plate, (b) Cylindrical stiffener and end-
closure plate, (c) Box structure, (d) Beam to column.
 5

OMNL4MVG 7719306

J L

Fig. 3. Typical Joint Designs that Suffer from Lamellar Tearing,

(a) Fillet welded T-joint. (b) Fu.i penetration T-joint. (c) Full-
penetration cruciform T-joint. (d) Full penetration corner joint.

1 2 1
Lamellar tearing has been found * ' * from within the lower
heat-affected zone (HAZ) (i.e., that region of the HAZ closest to the
unaffected base metal) to well into the base metal thickness. The
tearing may remain completely subsurface o; appear at plate edges or
1
at weld toes.
2
Considerable disagreement among investigators is reported
concerning the time and temperature of the onset of tearing. In
3 2
fabrication and testing situations, tearing has been reported to
occur shortly after passage of the weld pool when previous beads have
built up strain. The passing weld pool acts to cause: transient
2
overloading due to reduction in section support, heating regions
3 5
into the temperature range where ductility minimums ' can occur,
and additional strain due to thermal contraction. However, others
2
have reported the tearing to be a room-temperature delayed cracking
phenomenon.
 6

The steplike cracking that occurs has several distinct phases.

Figure 4 schematically represents these phases based on a description
by Farrar et al/' Duriag the initiation phase voids are usually formed
by decoherence or fracture of single elongated nonmetallic inclusions
or elongated clusters of them lying parallel to the rolling plane.
Additional void initiation mechanisms are reported such as splitting
2 7 2
along ferrite bands, ' liquation in reheated HAZ, intergranular
2 8 9
cracking, hydrogen assisted cracking *' and strain aging. However,
2 7-4
aost investigators ' believe that decoherence or cracking of
inclusions is the primary initiation mechanism. The initiated void
or tear, which is formed parallel to the rolling plane, is called
3
a "terrace." The second phase of lamellar tearing involves the
linkage or growth of terraces on common planes. The mechanisms of
2
terrace linkage that have been observed are necking, oicrovoid
coalescence, "zig-zag" tearing, intergranular cracking, quasi-cleavage,
and cleavage. However, the most commonly observed mechanisms are
2
necking and microvoid coalescence. The third phase of lamellar
3
tearing involves the formation of vertical shear walls to connect

ORNL-DWG 77-19306

PHASE 0 - LINKAGE OH GROWTH

OF TERRACES ON A COMMON PLANE

L A R G t p
CLOSELY SPACED SMALLER KlOpml INCLUSIONS LANAR »~3QMm> INCLUSIONS GIVING
ALSO GIVING EASY OEC0MESION ANO REQUIRING « PHASE I • EASY OECOHESION ANO REOUIRING LOW
LOW ENERGY FOR FORMATION OF TERRACES ENERGY FOR FORMATION OF TERRACES

Fig. 4. Schematic Representation of the Phases of Lamellar Tearing.

 7

regions of linked terraces on different planes. This third phase

produces the steplike appearance of the cracking, and it occurs by
2 5
ductile shear through regions of aicrovoids. »
5
Lamellar tearing itself has been reported, from a fracture
mechanics standpoint, as being a subcritical growth phase. Once the
crack reaches a critical size through lamellar tearing, brittle
fracture becomes the postcritical growth mechanism.
The foregoing description of lamellar tearing indicates that
a number of complex interrelated variables control the phenomenon.
Disregarding external load application, these variables can be
generally grouped into three categories: material, welding, and
design.

MATERIAL VARIABLES

The carbon and low-alloy steels that are susceptible to lamellar

tearing have low short transverse (ST) ductility. This low ST is
2
related to three variables: nonmetalllc inclusions, matrix properties,
and embrittlement. Each of these variables can be associated with
the previously described phases of lamellar tearing.
Most investigators agree that nonmetallic inclusions are primarily
responsible for the low ST ductility. In conventional steelmaking
10
practice (CON) the maximum sulfur requirement is above 0.025 wt X.
Depending upon charge or slagging the attained sulfur level can
normally range from 0.010 to 0.030 wt X. The dominant nonmetallic
8 10
inclusions are sulfides and oxides. ' The deoxidation practice
8 10
determines what type of sulfides and oxides ?re present. ' For
correlation to lamellar tearing, Farrar and Dolby' have classified
the deoxiriation practices into two categories: non-aluminum-treated
and aluminum-treated. Table 2 summarizes the effects of deoxidation
practice and inclusion types on lamellar tearing in CON steels.
Semikilled (<0.1 wt 7. Si) and fully killed (>0.1 wt X Si)
are the normal deoxidation practices for the non-aluminum-treated
steel grades of interest. In the semikilled and fully killed steels
 8

Table 2. Effects of Deoxidation Practices and Inclusion

Types on Laaellar Tearing in CON Steels

Deoxidation Inclusion
Effect on Lamellar Tearing
Practice Type

Non-Aluminum-Treated
Semi Killed, Silicates Become aore elongated than the
Fully Killed s u l f i d e s during hot r o l l i n g and
are primarily responsible for
reduction in ST d u c t i l i t y
Type I Manganese Dispersed globules are deformed
Sulfides to lozenge-shaped inclusions
during hot r o l l i n g (aspect
r a t i o 2-20, length 2-200 urn in
r o l l i n g direction) and general­
l y do not reduce d u c t i l i t y
u n l e s s there are l o c a l l y high
concentrations
Aluminum-Treated
Fully Killed Type I I Manganese Rod-shaped inclusions become
Sulfides highly elongated during hot
r o l l i n g (aspect r a t i o s up to
1000, lengths several am in
r o l l i n g direction) and are
primarily responsible for the
reduction in ST d u c t i l i t y

Alumina Although the inclusions are

not Individually elongated
during hot rolling, elongated
clusters or galaxies can
produce locally poor ST
ductility
Fully Killed with Type III Manganese Inclusions become highly
Excess Aluminum Sulfides elongated during hot rolling
and are primarily responsible
for the reduction in ST
ductility
Alumina Same as for fully killed
 9

the oxides are present as silicates and the sulfide is present as

2
type I manganese sulfide. The type I manganese sulfide normally
precipitates in the ingot as dispersed globular inclusions.* During
hot working this sulfide is deformed to lozenge-shaped inclusions
(aspect ratio 2-20, length 2-200 \m in rolling direction).' It is
believed' that oxygen present in these type I MnS inclusions provides
solid solution or dispersion strengthening, which prevents them from
being readily deformed during hot rolling. Because of the lozenge
shape and distribution, the type I manganese sulfide does not
generally reduce the ST ductility unless there are locally high
6
concentrations. Locally high concentrations can be caused by high
sulfur levels (*0.5 wt I) or segregation. The silicate inclusions
2 8
are reported * to b? principally responsible for reducing the ST
ductility because they become more elongated than the sulfide inclusions
during hot rolling. The more elongated silicate decoheres in preference
to lozenge-shaped sulfide because of greater stress intensification
and lower surface energy. Because the silicates are primarily
responsible for reducing the ST ductility, the sulfur level can
give an incorrect indication of susceptibility to lamellar tearing
in non-aluminum-treated steels.

Fully killed (>0.1 wt X Si, usually 0.02-0.08 wt % Al) and

fully killed with excess aluminum (>0.1 wt % Si, >0.06 wt '/. Al)
are the deoxidation practices for the aluminum-treated structural steel
grades of interest. The oxide is present as alumina clusters or galaxies,
and the manganese sulfide is present as type II for the fully killed and
2 8 10
type III for the fully killed with excess aluminum. * ' Depending upon
the specific practice, variations of manganese aluminum silicates and
2
aluminum silicates can also be present. The type II manganese sulfide
inclusions solidify in the ingot with a rod-shaped interdendritic
8 1 0
morphology. • * Under certain rolling conditions at low tempera­
tures, the type II sulfide can be deformed to give aspect ratios
8
of up to 1000 and lengths of several millimeters. The type III
9
sulfide inclusions solidify with an angular morphology. Because
of the high plasticity of the type III sulfide inclusions during
 10

hot rolling, they can become even more elongated than the type II.
The alumina galaxies are trapped by steel dendrites during solidifi­
10
cation. Although alumina inclusions are not individually deformed
during hot rolling, the groups or galaxies can be deformed into
elongated clusters depending upon their original spacing in the
2
ingot. However, it is the sulfides that are reported *' as primarily
responsible for the low ST ductility in the aluminum-treated steels.
An obvious question that can be raised at this point is whether
or not non-aluminum-treated and aluminum-treated grades differ in
susceptibility. In effect this question asks what is the difference
between the silicates, which are dominant in non-aluminum-treated,
and the sulfides, which are dominant in aluminum-treated. Oates
11
and Stout reported in 1973 that silicates were more detrimental
2
than sulfides. However, Ganesh and Stout in 1976 and surveys
1
corducted by the Welding Institute reported that the phenomenon
of lamellar tearing is too complex to simply relate susceptibility
12
to steel grade or inclusion type. Another study showed that the
crack opening displacement to cause tear initiation was independent
of inclusion type but depended instead upon projected lengths in
12
the rolling plane. However, this study also showed that lower
volume fractions of Type II manganese sulfide were required to
achieve high projected lengths.

During sufficient straining in the ST direction, either the

interface between the matrix and the elongated inclusions decoheres,
or the inclusion fractures to initiate a tear. This tear initiation
mechanism can be enhanced by differences in thermal expansion between
3
the inclusions and steel matrix. The sulfide inclusions, which
2
contract more than the steel matrix, can form a void at the interface,
thereby providing a sitt for the tear to Initiate. The silicates
3
and alumina, which contract less than the matrix, would form residual
stresses, which could provide a source of stored energy to aid tear
initiation. It is therefore readily apparent that nonmetallic
inclusions play a primary role in the void initiation phase of
lamellar tearing.
 11

The spacing between the elongated inclusions is an important

factor in the linkage of terraces on coanon planes. This spacing
depends upon both the overall inclusion content and any mechanisms
that tend to cause segregation. As the inclusion spacing decreases,
less energy will be required to tear the intervening matrix.
Therefore, the elongated inclusions also play an important role in
the terrace linkage phase of lamellar tearing.
Small, finely dispersed inclusions such as alumina are important
in both the terrace linkage and shear wall formation phases. These
inclusions provide sites for microvoid formation. During tearing,
microvoid coalescence is enabled to readily proceed.
The properties of the steel matrix can be important in all
2 8
phases of tearing. Ferrite-pearlite banding has been reported '
as causing both initiation and propagation of lamellar tears. This
is partially because the ferrite has a lower cleavage fracture
8
stress than the pearlite. In addition, the lower toughness of
5
the pearlite could provide a crack path. A previously described
model characterizes lamellar tearing as the subcritical crack mode
and brittle fracture as the postcritical mode. In terms of this
model the fracture toughness of the matrix is very important with
respect to tear propagation. The grain size of the matrix is an
important factor ..1 determining its fracture toughness. This is
because the grain size affects the ductile-brittle transition
temperature.

The thermal and strain cycles that result from welding can
harden the steel matrix and thereby enhance both crack initiation
and propagation capability. Crack initiation by splitting in a
coarse-grained HAZ has previously been described.
3
Strain aging has also been previously described as a mechanism
that could potentially embrittle the HAZ and reduce ductility. The
introduction of hydrogen during welding provides yet another mechanism
for assisting crack propagation.

f
t.
 12

WELDING VARIABLES

The principal effects of welding on lamellar tearing are strain

caused by restraint of thermal contraction and a temperature transient
caused by heat flow from the passing molten weld puddle. The
variables that interact with these effects can be categorized as
being related to the welding process or joint design. The joint
design variables will be discussed in the next section.
Several welding variables affect lamellar tearing susceptibility.
1 3 8 11 13
These are r e p o r t e d * ' » ' to be heat input, preheat, filler
material, and bead sequence. The effect of increasing heat input
13
or preheat is reported to improve the postweld ductility. Ganesh
13
and Stout report four possible factors for the improvement in
ductility due to high heat input:
1. increased penetration, which causes interception and blunting
of existing tears;
2. increased deposition, which decreases the required number of
weld beads and consequent strain cycles;
3. reduction in the cooling rate, which allows stress relaxation;
4. producing a softer and tougher HAZ.
1
An additional explanation is that higher heat input processes
produce lower strength weldments, which will accommodate more of
the strain. Although increased heat input has been reported to
13
decrease the susceptibility to failure, it increased the amount
of subcritical tearing. These subcritical tears were located mainly
13
in the coarsened H A Z . The location and increased amount of subcri­
tical tears might be detrimental during some service loading conditions.
In addition, the use of high heat input for highly restrained joints
3 1
could lead to more overall shrinkage. In a survey of fabrication
experiences, increasing heat input within a process was reported
as ineffective.

The factors cited above for improvement in ductility due to

11
high heat input also apply to preheating. In addition Oates and Stout
have reported that preheating can raise the temperature of certain
materials above their brittle fracture transition, thereby retarding
 13

early propagation of lamellar tearing. For material initially abo/e

its transition temperature, they reported a reduced effect from
preheat and attributed the small improvement in tearing susceptibility
1
to a reduction in thermal gradients. A survey of fabricators
reported no successes with preheat, and one case of increased tearing.
In general, when investigators discuss the effects of preheat, they
5 13
usually ' add a warning that it may increase contraction strains.
If either the recommended material or inspection requirements
described in a later section are used, restriction on the use
of preheat are not felt to be necessary.
The proper selection of filler metal is important so that the
strength of the base metal is not overmatched. If the weld metal
is significantly stronger than the base metal, none of the contraction
strain will be accommodated by the weld metal. This results in an
increased strain buildup in the base metal tilth a subsequent increase
in tearing susceptibility. Recommendations for limiting the strength
of the weld metal are discussed later.
The sequence of weld bead deposition can significantly affect
the amount of strain and how localized it is. The use of balanced
bead deposition will keep the strain more symmetrical thereby reducing
localized buildup. The alternation of welding direction after each
13
pass has also been reported to reduce initiation of lamellar
tearing. The reason for this can be related to the initiation of
tearing in multipass welds from a site where straining has been
localized during earlier passes. Reversing the welding direction
after each pass can reduce the localization of straining.
In addition to the welding variables described above, several
techniques have been used to modify the thermo-mechanlcal effects
of welding. These techniques are buttering, peening, and stress-
relief heat treatment. In the buttering technique, the weld is made
on top of a deposit of filler metal that has been previously applied
to the plate and ultrasonically examined. The filler metal deposit
consists of one or more layers, which are applied directly to the
plate surface or in a gouged out area. Figure 5 (based on ref. .'.)
 u

•UTTIHIDLATII*

is TO n -
i TO 'n «

Fig. 5. Welds Hade on a Buttered Layer (a) Applied Directly to

Plate Surface and (b) Applied in a Gouged-Out Area.

1
illustrates welds made on a buttered layer. The recommended dimensions
for the buttering layer(s) are that it should extend 15 to 25 mm
beyond each weld toe and be 5 to 10 mm thick. The purpose of the
buttered layer(s) is to accommodate the weld contraction strains
and displace the HAZ from the susceptible plate. As a result, the
weld metal used for buttering should not overmatch the strength of
the plate. Also, it is important to avoid introducing hydrogen
11
and to regulate heat input to produce sound buttering. * Although
1 lh
the buttering technique has been reported ' to be very successful
in avoiding lamellar tearing, a few instances of tearing resulting
from battering directly on the plate surface [Fig. 5(a)] have been
1 15
reported. In addition, one study reported that single buttering
layers did not prevent lamellar tearing in a prototypic box section
corner joint fabricated from susceptible plate. In general, if the
proper controls are placed on the buttering procedure and an adequate
size buttered layer is used, this technique should be quite successful
in preventing lamellar tearing. However, the cost for using this
 15

technique, the additional tiae required, and the chance of error

introduced by any additional fabrication step must be weighed
against buying improved steel.
Peening consists of mechanically working the weld beads by
hammer blows or shot impingement to reduce residual stress, and
thereby lessen the risk of lamellar tearing. Although peening
has been used in the welding industry for a number of years in
applications not related to lamellar tearing (prevention of distortion),
the technique is not easily controlled and therefore is subject to
1
a wide variation in results. In addition, it is reported ' to
produce adverse effects such as reduction in toughness under certain
1
conditions. A survey of fabricators indicated that peening reduced
tearing in only one out of eleven reported cases. Although ASME
17
Code Sect. Ill Division 1, Subsection NF currently allows the
use of peening, it does not appear to be a viable method for reducing
lamellar tearing. However, peening of weld metal should not worsen
the risk of tearing in susceptible material and therefore need not
be prohibited on the basis of lamellar rearing prevention.

Intermediate heat treatment is another technique that has

been used to reduce residual stresses. For large components,
which would require localized heating, this technique might increase
the overall contraction strains. In addition, this heat treatment
might cause some decoherence at inclusion-matrix interfaces and
thereby increase the amount of indications seen by ultrasonic
1
examination (UT).' In general the technique has not been reported
to be particularly successful.

DESIGN VARIABLES

The design variables that affect lamellar tearing are related

to strain accumulation and concentration in the through-thickness
direction. These variables can be divided into two categories:
joint design and fabrication sequence. Three major factors in
joint design affect the accumulation and concentration of
 16

through-thickness strain. These are weld orientation, balance, and

size. The orientation of the weld is obviously an important factor.
If a weld is oriented such that it does not cause through-thickness
contraction strains, then the likelihood of laaellar tearing should
be eliminated. Figure 6, based on refs. 1 and 4, illustrates methods
for reorienting Che weld to reduce through-thickness contraction.
In some T-joints, the weld can be reoriented by using a single
electro-slag pass [Fig. 6(a)]. In corner joints the weld can be
reoriented by placing the bevel on the plate that would have been
through-thickness loaded [Fig. 6(b)). Redesign of complex structural
connections can often reorient the weld to avoid through-thickness
11
strain [Fig. 6(c)].
The balance or symmetry of a joint design is an important
factor controlling the concentration of strain. Joint designs that
deposit the weld from, one side cause the strains to concentrate and
1
accumulate on that side. Figure 7 illustrates methods for balancing
joint symmetry to reduce the concentration and accumulation of strain.
The size of the weld is an important factor in joint design,
which influences the amount of strain accumulation. In multipass
welds the size of the weld determines how many passes are required
to fill the joint. The number of passes subsequently influences
the amount of strain accumulation. Wide groove angles, unnecessary
use of full penetration welds instead of simple fillets, and increasing
the voluaie of weld metal to reduce the sectional loading stress
are coranon occurrences that Increase the size. Recommendation?
for requirements to limit the weld size are discussed in a later
section.
The fabrication sequence is an important aspect of design, which
can have a significant effect on strain concentration in a particular
joint. The fabrication sequence determines the amount of restraint
that each joint in a multljoint component will experience. This
restraint subsequently influences the amount of strain. If the more
susceptible joints in a component can be made initially or as
separate subassemblies where restraint levels are lower, weld contraction
strains will be allowed to dissipate.
 17

ORNL-DWG 77-19308

SUSCEPTIBLE JOINT IMTROVEO JOINT

la)

IM
t>>

Fig. 6. Methods for Reorienting the Weld to Reduce Through-Thickness

Contraction Strains, (a) Reorienting the weld in a T-joint by using a
single pass electroslag weld, (b) Reorienting the weld in a corner
joint by placing a bevel on the susceptible plate, (c) Redesign of a
complex structural connection to reorient the weld.
 18

ORNLOWG 7749309

SUSCEPTIBLE JOINT IMPROVED JOINT

Fig. 7. Balancing Joint Symmetry to Reduce the Concentration

and Accumulation of Strain.

TESTS FOR ASSESSING MATERIAL SUSCEPTIBILITY

The ideal test for assessing susceptibility to lamellar tearing

must be suitable for production quality control and incorporate the
actual joint design, restraint, welding procedure, and material to be
welded. However, in practice, while it might he possible to use the
actual joint design and welding procedure, and to some degree the
restraint, the material sample that is tested may not be representative
of the actual area to be welded. This latter situation is due to
12
the heterogeneous nature of ST ductility, which has been reported
as varying between ingot yields of the same steel grade, between
plates made from different locations within an ingot, and between
different locations within the same plate. Unfortunately this
situation can manifest itself In steels that have a low overall
inclusion content yet have areas of localized segregation. Because
of this situation, any destructive test inherently can incorrectly
characterize the material whether the characterization be too optimistic
or pessimistic. The use of a nondestructive examination (NDE)
 19

technique would then appear to offer the best solution since it

would enable surveying an entire plate or just the specific areas
to be welded.
3
Jubb has grouped the tests for assessing material susceptibility
into the following four categories:
1. tests without simulated or actual welding cycle,
2. tests with simulated temperature and strain cycle,
3. tests with temperature and strain cycles by welding, and
4. tests of a prototypic welded joint.
The dominant variable in these categories is the temperature and
strain cycle effects of welding. Although it is desirable to be
able to test material without imposing the thermo-mechanical effects
of welding, there is considerable controversy concerning the validity
if it is not done.

Tests Without Simulated or Actual Welding Cycle

The tests without a simulated or actual welding cycle can be

either destructive or nondestructive. Most of the reported work
concerning the use of nondestructive tests has been performed at
the Welding Institute. Because of the potential benefit of a
nondestructive technique, the Welding Institute devoted a considerable
amount of their work on lamellar tearing to developing an ultrasonic
18
test. The result of their initial three-year group-sponsored
project was the development of a 5-MHz unfocused probe, which
usad an integrated echo technique to examine discrete levels of
10
thicknesses under both plate surfaces. The integrated echo technique
measures the area under the reflections (a function of the size
and number of reflections) and uses it to represent the volume
fraction of inclusions. A short-pulse-length probe was used to
correct for interface echo masking of the layer under the surface
being scanned. A normal-pulse-length probe was used for the surface
layer opposite the side being scanned, with correction for signal
attenuation from intermediate inclusion layers. Corrections were
also made for variations in surface finish. Because of the differences
 20

in predominant inclusion types, separate calibration curves were

developed to relate integrated echo amplitude to short transverse
percentage reduction of area (ST Z RA) for aluminum-treated and
1
non-aluminum-treated steels. Wide-scale evaluation ' of the
technique discovered that incorrectly high values of ST Z RA
were being predicted because of variations in inclusion population
and inclusion-steel matrix interface (resulting from different
thermal and mechanical treatments) for different grades of steel.
19
From these results it was concluded that the development of a
technique applicable to a wide range of steel types was not probable.
19
Additional work at the Welding Institute on the UT technique
has been aimed at simplification of the procedure to eliminate the
need for complex calibrations. The need for surface finish correc­
tions was eliminated by using one standard finish. Corrections
for attenuation of reflections from the back wall were eliminated
by only scanning the surface layer from one side. Instead of
attempting to predict ST Z RA, direct comparison is made with
19
standards. Although some success has been reported with this
simplified technique, it is presently limited to a narrow rarge
of steel types, or to predicting relative variations in a given
plate.
20
However, a fabricator in Italy recently correlated ST Z RA
and ultrasonic test results. In addition, specifications for
specific steels to be used in North Sea offshore structures are
using both ultrasonic inspection and ST Z RA as susceptibility
21
tests. Even with these reported successes, it still must be
realized that Che application of ultrasonic examination to the
determination of ST Z RA in a wide range of materials will be
difficult.
Several destructive tests are performed without a simulated
or actual welding cycle. While some of these tests require actach-
ment welds for gripping or support, the welds are not meant to
provide temperature and strain cycles. The more commonly reported
tests are tensile, slice bend, Charpy V-notch (CVN), and analysis
of inclusion population.
 21

The ST tensile test is currently the most controversial test

for assessing susceptibility. The controversy is because the
test offers a standardized method readily applicable to quality
control which makes it highly desirable, yet its variability and
correlation with lamellar tearing are disputed. The use of the
12 21
ST tensile test is reported ' as already being an integral
part of steel specifications for critical applications such as
21 22
offshore structures. In addition, it is reported ' that this
test has been included in recently issued experimental standards
in France and Germany. The ST tensile property that is most
frequently used is Z RA. The X RA is considered 23
to provide a
good indication of ST ductility associated with a fracture propagat­
11
ing parallel to the rolling plane. It is also reported to be
the tensile property that correlates most closely with tearing
susceptibility. In addition, it can be more easily standardized
than elongation since gage length will vary with plate thickness.
The validity of the ST tensile test is usually questioned because
3 2
of the following: ' '*

1. only a small cross-sectional area of material is tested, resulting

in a statistically small sampling;
2. the fracture of the test specimen usually occurs near the
center of the plate, while the surface region is where lamellar
tearing is reported;
3. the results can be greatly influenced by single large inclusions
or clusters, while neglecting the steel matrix properties;
4. there are conflicting reports as to the correlation between
ST % RA and susceptibility to lamellar tearing.
The concern about the statistically small area sampled is related
12
to the reported heterogeneous nature of ST ductility. Any of the
destructive tests that rely on a sampling plan are subject to this
error. Although some tests may evaluate a larger area, the total
area tested will always be a small fraction of the material it
represents. One philosophy that can be used to justify this test
is that if a sufficiently high minimum ST ductility acceptance limit
 22

is used, then the poor areas that may escape testing should be
saall and localized and therefore not create significant tearing.
12
This philosophy is supported by the detailed investigation of
ST Z RA of the plate yields of two production ingots of fully silicon-
killed carbon—Manganese steel. Figure 8 shows the location and
12
dimensions of the plates produced froa the two production ingots.
The testing consisted of 23 sets of tests froa ingot A plates and
21 froa ingot B plates with six .totalized specimens per set. Figure
12
9 shows the reported test locations and the ST Z RA values. The
ST Z RA values of the ingot A plates were higher and aore consistent
while those for the ingot B plates varied considerably. If the
2 1
conservative recommended ' ' minimum ST Z RA acceptance limit of
25Z is used, then none of the sets from ingot B would be acceptable,
while seven sets from ingot A would have been rejected. This
indicates that the use of a conservative minimum acceptance limit
can screen out plates with low and widely varying ductilities. It
also shows that the limits can be overly conservative.

oomionc " <93io

MCOT A FlATfS

Fig. 8. Location and Dimensions of the Plates Produced from Two

Production Ingots.
 23

moor * M H J MGOT • MTH

•S rr -n

-» "S » '5

Fig. 9. Location of Tensile Tests and ST Z TA Values, Mean (Range),

for the Plates Produced from the Two Production ingots.

The location of the ST tensile test has been a concern because it

does not test the surface area, where tearing normally occurs. Oates
and Stout reported 11
significant differences in ST Z RA (average of
19.4 vs 2.8Z) between specimens with gage lengths centered on plate
mid-thicknesses vs 6.4 mm from the surface of 64-mm plate. This
could be because surface regions are reduced more during hot rolling
and the inclusions located there could be more elongated. However,
the restraint of plastic deformation by the hardened HAZ of the
weld attaching the testing grips must also be considered. One
9
investigation has already shown that when the plate thickness is
below 30 mm, ST % RA begins to decrease from restraint of plastic
deformation by the KAZ of the grip attachment welds. The use of
friction welding for grip attachment produces a small HAZ and is
 24

9
the recommended procedure. In addition, heat treatment of the
attachment welds has been reported as a mean;- of eliminating the
restraint from the HAZ.' However, this technique would also heat-
treat the base metal and alter its properties. Lamellar tearing
occurs near the plate surface because of localized thermo-mechanlcal
effects of the weld cycle. More data need to be developed to indicate
whether the surface material is more susceptible.
12
Large single inclusions or clusters have been reported to
reduce the ST Z RA. These decreased ST Z RA values do not necessarily
account for the resistance to further tearing that the steel matrix
raay have provided. The effect of the large Inclusions or clusters
can therefore be viewed as making the test more conservative.
5 23 25
Several studies »'» » have shown a correlation between ST Z RA
and susceptibility to tearing and have made recommendations for
11
acceptance limits. However, another investigation has reported
23
an apparent lack of correlation. In one study, 15 steels with
known fabrication history of tearing and 20 with known fabrication
success were evaluated to determine any correlation between ST Z RA,
--tearing history as a result of fabrication, and type of structure
(i.e., degree of restraint). The results of this study were three
acceptance limits for ST Z RA based upon joint restraint. These
minimum ST X RA acceptance limits are 25Z for high restraint such
as nozzle penetrations, 15Z for moderate restraint such as box
and cruciform joints, and 10Z for light restraint such as fabrica­
tion of I-beams. In addition, the following points were made:

1. the acceptance limits are appropriate to yields strengths

below 400 MPa (58 ksi);
2. specimen extraction should be from the roost critical fabrication
region or from a number of locations;
3. at least six specimens should be tested from any area being
evaluated;
4. the test is prone to large scatter. Areas having a wide range of
ST X RA values indicate low risk of tearing. Only consistently low
values, with a low mean and narrow range, indicate a high risk of
tearing.
 25

Another study correlated tearing susceptibility of 40 steels

with ST Z RA, as ranked by five residual stress levels in the
26
Welding Institute Window Test. The study showed a good correlation
between susceptibility and ST Z RA, with the same scatter trends
23
as previously reported (i.e., scatter increasing as susceptibility
decreases). As a result of this study the following minimum ST Z RA
acceptance limits were recommended: 25Z for high restraint, 15Z
15
for moderate restraint, and 8Z for low restraint. One investigation
consisted of correlating tearing in three steels (with varying ST Z RA
and fabrication histories) during fabrication of five types of test
structures with various levels of restraint. The result of this
study again was a set of minimum ST Z RA acceptance limits, which
were: 20Z for highly restrained, 15Z for moderately restrained,
and 10Z for lightly restrained. It was also noted that the possible
error for measurement for ST Z RA was high in the range 0-202. In
5
a final study, the incidence of tearing in three large weldment
assemblies (representative of pressure vessel fabrications) was
compared with the ST Z RA of the ASTM A 515 grade 65 steel plates
used. The result of the investigation was a guideline of 16Z
minimum ST Z RA for the specific fabrications and material type used.
1
However, in one investigation,' which compared ST Z RA with the
critical amount of weld restraint needed to cause tearing during
the Lehigh Lamellar Tearing Test, no apparent correlation was found.
21 25
The slice bend test ** is another destructive test that is
performed without the thermo-mechanical effects of the weld cycle.
The test was developed to provide a means of assessing tearing
susceptibility suitable for plant quality control without some
of the disadvantages of the ST tensile test. Although the slice
bend test offers some improvements in t^w.ing technique, it is
still susceptible to the sampling problems associated with all
75
destructive tests. The original test procedure has been modified.
2
It is currently * performed by attaching a thin slice of plate to
a backing bar and then subjecting the composite to three-point
bending such that the through-thickness direction of the slice is
in tension. Figure 10 shows the manufacturing and testing sequence"'*
 26

GBNLOV.G nj'36:

Fig. 10. Manufacturing and Testing Sequence for the Slice Bend
Test, (a) Removal of slice from plate, (b) Attachment of slice to
backing bar. (c) Three-point bending of the composite to produce
through-thickness strain in the slice.

for the slice bend test. The heavy backing bar enables uniform
straining across the thickness, and the amount of strain imposed
is controlled by various measuring techniques, which were still
being investigated. The criterion for assessing tearing suscepti­
bility is the measurement of total tear length that has progressed
through the linkage or shear wall stages for a given testing strain.
By relating the testing strain to joint restraint levels, a maximum
tear length acceptance limit can be established for various testing
strains. Although the slice bend test requires more work on strain
measurement technique and establishment of tear length acceptance
2
criteria, the following advantages " indicate that it should be
given further consideration:

1. it is a simple test, which can be readily performed on a plant

quality control level;
2. the test evaluates the complete thickness;
 27

3. the test produces realistic lamellar tearing anti assesses

the important linkage and shear wall stages, thereby enabling
it to be more sensitive and possibly reduce the overconservatism
of the ST tensile test.
The ST Charpy V-notch (CVN) test is another destructive test
that is usually performed without the thermo-mechanical effects of
a weld cycle. However, attachments must be welded for testing
thin plate or locating the notch near the surface. This test is
highly standardized and readily usable for plant quality control.
1
There has been some controversy concerning the applicability of
the CVN test, since it has a higher strain rate than that associated
with lamellar tearing. In addition, because the test only evaluates
the plane parallel to the notch, it does not necessarily test the
5 11
most susceptible region. Two studies report ' a correlation
between tearing susceptibility and CVN tests. One study correlated
5
CVN tests with prototype weldments while the other utilized the
11
Lehigh Lamellar Tearing Test. The result of the prototypic weldment
study was the proposal of an acceptance limit for the specific
9
joint types and material that were evaluated. However, another study,
which evaluated susceptibility by the Welding Institute Window Test,
concluded that there was no definite correlation. The CVN test
appears to be more applicable to characterizing the linkage or
shear wall stages of tearing or the brittle fracture mode when the
lamellar tear has reached a critical size. Since this test will be
sensitive to changes in matrix properties, it would appear to be
a good test to run in conjunction with ST tensile tests, as already
5
suggested. The main problem is that there are no widely applicable
acceptance limits available.

The analysis of inclusion population by quantitative metallographic

examination is another testing method that is performed without imposing
the effects of a welding cycle. The parameters that have been used
to characterize the inclusion population are: number, volume fraction,
length (parallel to rolling plane), aspect ratio, mean free path, and
projected length. Because of the small area scanned per sample, this
 28

method would probably require an extensive saapllng plan. The

high costs of autoaatic scanning equipment and large number of
tests required do not sake this testing aethod readily applicable
as a plant quality control tool. In addition, there is soae contro­
versy concerning the correlation between inclusion population analysis
and susceptibility to tearing. Generally, the correlation between
susceptibility and number or volume fraction of inclusions has been
3 12
reported ' to be weak. Correlation is improved by using factors
12
related to inclusion length, as evidenced by the reported correlation
between ST X RA and maximum inclusion length. The inclusion projected
length, which is the total length of inclusions parallel to the
12
rolling plane per unit area, is reported to relate several in­
clusion population variables to the direction of fracture propagation.
12
The projected length has shown good correlation with notched bend test
maximum crack opening displacement (COD). However, in another
2
investigation, the inclusion analysis did not provide a good correla­
tion. The possible explanation reported for this poor correlation
was that the quantitative analysis did not account for smaller inclu­
sions or matrix properties that affect the propagation of tears.

Tests With Simulated Temperature and Strain Cycles

Simulated weld temperature and strain cycles can be imposed on

specimens that have been sectioned from the material to be evaluated.
Equipment such as the Gleeble is currently available and capable of
providing a simulated HAZ with the proper orientation to the through-
thickness direction. However, this type of testing is more suitable
for laboratory investigations than for production quality control.
9
One investigation has reported using this weld cycle simulation
technique and then performing tensile tests. The results of this
testing showed a decrease in the ratio of ST to longitudinal direction
RA.
 ->o

Tests With Temperature And Strain Cycle By Welding

Two categories of tests within this group use welds. Methods

such as the Cranfield Test, Welding Institute Window Test, and
Lehigh Lamellar Tear Test use a weld to make a direct assessment
of the tearing that results. In contrast, the ST notched bend
test utilizes the resultant HAZ from a weld to enable COO testing
in the region where tearing normally occurs. Tt testin methods
0

in this group are still susceptible to the same sampling errors

affecting all destructive tests.
Although the ST notched bend test can provide COD measure­
12 27
ments of unaffected base metal, the reported s t u d i e s * have
1Z 27
examined the HAZ of welds. In these ST notched bend t e s t s *
full plate thickness specimens are removed and attachments are
welded to the two surfaces parallel to the rolling plane. Figure
11, based on ref. 27, shows the manufacturing and testing sequence
for the specimens. One weld is a square butt made by electron beam

Fig. 11. Manufacturing and Testing Sequence for ST Notched Bend

Specimens, (a) Assembly of components for composite specimen, (b)
Welding attachments, (c) Machined specimen, (d) Testing of specimen.
 30

and is meant only to attach an extension block [Fig. 11(b)]. The

other veld is a 45* bevel-groove made by the manual shielded metal-arc
(SNA) process to attach the stem plate [Fig. 11(b)]. The multipass
SMA weld is used to impose through-thickness strains and thermal
cycles. After the welds are made, specimens are machined from the
composite, and a fatigue cracked or V-shaped notch is oriented
parallel to the SMA weld, at various locations in the vicinity of
the HAZ (Fig. 11(c)]. During three-point bending, COD measurements
are made as crack initiation and propagation occur on a plane parallel
to the plate surface [Fig. 11(d)]. The COD to maximum load (6 max)
is normally used as the assessment parameter because other values,
such as the area under the load vs displacement curve, are too
27 12
dependent upon test machine stiffness. In one investigation,
quantitative metallographic analysis of inclusions was performed
on samples adjacent to the notch bend test specimens. As previously
reported, a correlation was demonstrated between the mean 6 max
and the inclusion projected length. This correlation was shown to
27
be independent of the type of inclusion. However, attempts to
correlate notched bend test results with fabrication history or
ST Z RA have shown discrepancies because the notch location did
not coincide with the most susceptible area. The notch bend test
27
has been characterized as unsuitable for production quality control
because it is a complex and expensive test to run.
28
The Cranfield Test consists of a stem plate and a section of
base metal to be tested. Figure 12 shows the details of the Cranfield
Test specimen, adapted from ref. 28. The end of the stem plate is
beveled such that it forms a 45 or 60* Vee when seated on the surface
of the base metal plate. Multipass welding of the VEE, with interpass
cooling to allow full thermal contraction, results in a large strain
concentration at the root of the Vee. This strain is imposed on the
through-thickness direction of the base metal plate and results in
lamellar tearing under the root of the Vee when the material is
susceptible. The amount of tearing is determined by breaking off the
stem plate and measuring the percentage of heat-tinted fracture area.
26
Correlation of test results with fabrication history has been poor.
 31

Fig. 12. Details of Cranfield Test Specimen.

3
Because the test conditions are so severe, it can only identify highly
resistant materials. In addition, variability of testing conditions
can make it difficult to apply uniform strain cycles in different tests.
8
One investigation reported that, because cracking propagated only in the
HAZ, poor correlation was obtained when inclusions were outside the
HAZ. The use of the Cranfield Test for production quality control
would be difficult because of the variability of testing conditions
and the severity of the test. No recent investigations have reported
success with this test for assessing susceptibility.
26
In the Welding Institute Window Test, a plate of the base material
to be tested is inserted through a rectangular hole or window in a
restraining plate to form a cruciform joint. Figure 13 shows the
details of the test procedure, adapted from ref. 26. The window
sides adjacent to the two test plate surfaces parallel to the rolling
plane have a Vee bevel. Single-pass, full-penetration fillet welds
are initially made in two grooves on one side of the test plate to
anchor it. Similar test welds are then made in the two grooves on
 32

PtSTKAIMT »ukTt

- TIST »LATt

TtST tttLDS

ANCHOR WCLOS

— LOCATION Of
LAMtLLAR U A R
RiSTRAINT FLATt

Fig. 13. Details of the Welding Institute Window Test.

the opposite side. The restraint provided by the initial anchor welds
promotes the occurrence of lamellar tearing in the test plate during
the welding of the second side. Visual or metallographic observation
9
cf tearing is the criterion for failure. One investigation has
reported the following difficulties with the test:
1. root cracking can induce lamellar tearing and therefore confuse the
test results;
2. since the amount of restraint varies as a function of the restraining
plate and test plate thicknesses, it is difficult to assess test
results quantitatively.
9
Although this same study has demonstrated methods to lessen these
problems, they are too complex and require carefully reproduced welding
procedures. As a result of these problems, the test is not readily
applicable to production quality control.
11
The Lehigh Lamellar Tear Test was developed to enable quantitative
evaluation of the variables affecting lamellar tearing. The test is
 33

11
described by its developers as suitable for laboratory use. Figure 14
shows the details of the test, adapted from ref. 2. A cantilever
beaa is welded to the surface of a rigid vertical test plate. Throughout
the test, an external load is applied to the cantilever to produce
a constant through-thickness restraint on the test plate. Because
a multipass weld is used, the load necessary to maintain constant
restraint changes after the deposition of each layer. This variable
load must be calculated from load lever arm length, width of test
plate, thickness of cantilever beam, and the thickness of weld after
the deposition of each layer. The criterion for assessing the suscepti­
bility to lamellar tearing is the determination of thi critical weld
restraint level (CtfRL) just necessary to cause tearing. The CWRL
is determined by testing at various increasing restraint levels until
tearing is detected visually, metallographically, or ultrasonically.
2
The test has been used to examine a wide range of steel types and
the material variables affecting tearing. Modification of the tearing
susceptibility assessment criteria was necessary for evaluating welding
13
variables. The test had to be instrumented to enable recording load
vs joint displacement. The modified susceptibility criterion was a

V
MULTIPASS * E I D

m
CANTILEVER »LATE

L O C A T I O N O * LAMELLAR TEAR

RESTRAINT H A T E

Fig. 14. Details of the Lehigh Lamellar Tear Test.

 34

postveld energy index (Ep), which was calculated from the product of
fracture stress and displacement required to cause failure. Although
l l3
the Lehigh Lamellar Tear Test has demonstrated »* ' its usefulness
2

as a laboratory tool to study pertinent variables, it does not appear

to be readily applicable tor production quality control. This is
due to both the complexity of the testing procedure and the absence
of established acceptance criteria correlated with fabrication history.

Tests of a Prototypic Welded Joint

The results from testing material in a prototypic welded joint

are often very difficult to correlate with actual production situations.
This is mainly due to the difficulty of duplicating the exact welding
conditions and restraint that will be present during production. In
addition, as with all destructive tests, the material that is tested
may not accurately represent the piece from which it was removed.
Because of the potential variability of welding conditions and restraint,
the test inherently lacks the ability to quantitatively assess tearing
susceptibility. Jubb* has concluded that although prototypic weldments
might be useful for investigating certain parameters, they cannot
be recommended for general assessment of tearing susceptibility.
This testing method is also undesirable for production quality control
because of its complexity and expense. Although several recent
5 15
investigations ' have used prototypic weldments to study tearing
variables or other test methods, none have proposed its use as a
material assessment test.

DETECTION AND REPAIR OF LAMELLAR TEARS AFTER WELDING

Because lamellar tears can remain completely subsurface, such

NDE Techniques as visual, dye penetrant, and magnetic particle inspec­
tion cannot be used. Radiographic examination cannot be used both
1
because the Inclusions can mask defects and because the orientation
of the tearing makes detection difficult. Ultrasonic testing is the
normal method used to inspect for lamellar tears. However, the
 1 11
interpretation of UT results can be a problem. " Cases of misinter­
1
preted UT indications leading to unnecessary repairs have been reported.
Soaetiaes an angled shear-wave UT examination is needed to detect
the presence of the shear walls, which are characteristic of lamellar
21
tearing. Improper equipment gain settings and acceptance criteria
can also cause clusters of Inclusions or microstruetural bands to
1
be misinterpreted as tears. Unfortunately, the materials that are
more susceptible to lamellar tearing contain the type of defects that
make UT difficult. For example, thick d<rty plates can attenuate
ultrasound to the extent that interpretation may be difficult. Recent
22
work at the Welding Institute provides an example of this situation.
Ultrasonic testing of sections from a box-header that failed during
fabrication indicated tearing present in several locations. However,
netallographic examination revealed that inclusions had decohered
from the matrix, but had not linked to form tears. During tensile
testing of these sections, failure occurred on other planes. The
above discussion indicates uncertainty in the interpretation of
results of UT of production weldments. In addition, reliance on
detection by UT to avoid lamellar tearing nay be a poor substitute
for selection of improved material.

Ultrasonic examination of finished welds or buttered layers

does provide a conservative approach. The techniques arc sufficiently
sensitive such that rejectable tears should not be missed, but instead
it is more likely that acceptable indications may be rejected,
requiring subsequent repair. When the fabricator decides to use
ultrasonic examination instead of selecting material with improved
ST ductility, he will be risking repair costs but yet not jeopardizing
reliability.
The repair of lamellar tears can be very difficult and costly.
1
One study * indicates that in the repair of highly restrained connections,
the repair can be more detrimental than the original weld. This is
because at the time of the repair there is usually a larger amount
of restraint. The additional mechanical and thermal cycles from
repair welding can also simply cause tearing to occur at a greater
 36

depth in the plate thickness. A good exaaple of this situation was

2
provided by metallographic examination ' of a T-butt joint used to
attach a circumferential stiffener plate to a cylinder. This joint
had experienced extensive tearing during fabrication and was
subsequently scrapped. The examination revealed that an attempted
repair had caused further tearing below the repair weld.
A successful repair often requires drastic procedures. These
procedures might require complete disassembly of the welded joint,
grinding or gouging out the joint area and applying buttered layers,
and rewelding the joint over the buttering. This type of procedure
will of course be very expensive and cause considerable delay. From
an economic and schedular viewpoint, it may be more feasible in some
situations to scrap a component rather than try to repair it. Because
of the potential problems with the repair of lamellar tears, it would
appear to be economically justifiable to pay extra for improved
material properties rather than risk tearing.

SIGNIFICANCE OF SERVICE LOADING OF MATERIAL

SUSCEPTIBLE TO LAMELLAR TEARING

In order to evaluate the significance of lamellar tearing, the

effects of service conditions on both existent tears and tear-free
material must be considered. Generally, three types of service
loadings should be considered: static, dynamic, and fatigue.
Although most Investigators report only a few known cases of lamellar
tearing being initiated or propagated during service loading, critical
applications such as nuclear power plants and offshore structures
require Increased reliability because of stringent service conditions
ano the consequences of failure.
The effect of static loading on tear-free material, within design
limits, must be considered negligible when compared with the highly
localized loading produced by the welding cycle. It has been reported
that the localized strains under a weld can be as high as 2% without
even considering stress concentration. Because through-thickness
design stresses are limited such that they are significantly below
 37

the yield strength of the material,* strains experienced during

service will be significantly less than 0.22 (0.2Z offset strain is
normally the basis fcr defining the yield point of aost structural
materials). Therefore the strain that results from welding can be
at least 10 tiaes chat caused by service loading. A current group-
2
spcosored project ' at che Welding Institute is investigating the
significance of lamellar tearing under static loading. Although
this project is still in its early stages with only preliminary
results available, a few general trends have been detected. A portion
of che work completed involved tensile testing of cruciform test pieces
sectioned from a component that had been scrapped because of extensive
tearir-g during fabrication. The amount of lamellar tearing present
under tan fillet welds of the test pieces varied from none to extensive
tearing under bcth. The results of the tensile testing showed that
in most specimens the yf.pld strength of the base material was exceeded
before failure despite the presence of extensive tearing. In the
te3t piece with no tearing and in one with tearing under only one weld,
failure occurred by necking of the base metal rather than by extension
cf existing tears, as occurred in the remainder of tests. As would
be expected, the failure stress tended to decrease as the amount of
existing tearing increased. Although these preliminary results are
very limited, they do address the two situations chat must be consio^ed:
the effects of static loading on existing tears and tear-free material.
From these results and the previous discussion of design limits, the
welding appears more likely to cause tearing than static service
loads. In addition, loads above the yield strength of the base
metal are apparently required to propagate tears to failure even
though extensive tearing is present initially. However, because
the failure stress decreases as the amount of initial tearing
increases, failures could occur within design limits if the extent
of initial tearing is sufficiently great.

*ASME Code Section III Division 1, Subsection NF limits the design

stress intensity (3^) to a maximum of 1/3 of the yield strength for
class 1 components and 5/3 for classes 2 and 3, and for through-
thickness loading the Hesign stress is limited to 0.5.'„,.
 38

The significance of fatigue loading with respect to lamellar

tearing has received attention recently because of critical applica­
30
tions such as offshore structures. In one study, a heat of ABS
(American Bureau of Shipping) Grade EH 32 normalized steel with above
average static ST properties (ST X RA was greater than 20Z) intended
for use in an offshore semisubmersible drill rig was tested in fully
reversed, axial load-controlled fatigue. Although attachment welds
were used to fabricate the ST fatigue test specimens, no lamellar
tearing was detected by UT, visual inspection, or metallographic
examination of selected specimens. This testing is therefore applicable
to the effects of fatigue loading on material that is initially
tear-free. The results of the testing showed that the ST direction
had several times shorter fatigue life and 20 to 30Z lower fatigue
strength than the longitudinal direction. These differences were
s
greatest for high-cycle fatigue (greater than 1 0 cycles), where the
sensitivity to inclusion content, distribution, size, and shape
increases. In the ST direction the fatigue mechanism displayed several
rupture ..odes, consisting of quasi-lamellar tearing (decohesion at
inclusion interfaces, tearing, and cleavage) and fatigue striation.
However, ductile striation-type rupture dominated the longitudinal
direction. Inclusions both initiated and propagated fatigue cracks
in the ST direction, while in the longitudinal direction they were
insignificant.
31
Th< Welding Institute
; studied fatigue crack propagation in
tear-free material and weldments containing lamellar tears. Four
steels with varying susceptibility to lamellar tearing were tested.
The tear-free material was investigated with axially loaded, center-
notched plate specimens and edge-notched bend specimens, while sections
of a cruciform joint were used to study the weldments. In testing
of the tear-free material, the steel with the lowest ST % RA had
30
the greatest fatigue crack propagation rate. As previously reported,
the rupture mode consisted of both fatigue striation and quasi-lamellar
tearing. Increases in the amount of quasi-lamellar tearing mode
caused the crack propagation rate to increase. However, the results
 39

also indicate that a material must have extremely poor ST properties

before the fatigue crack propagation rate is seriously affected.
The evaluation of fatigue loading on existing lamellar tears was
limited to one steel because tearing did not occur in the cruciform
joints of the other three. The amount of tearing present in the
joints from the one steel varied from zero to 20% of the load-carrying
area. The results of the testing indicated that the fatigue strength
did not change as the amount of initial tearing increased. However,
this was attributed to the exceptionally low ST properties of the
steel. During the testing of the steels that did not contain tears,
all the failures initiated at the weld toe. However, on the basis
of the initial crack propagation studies, it was predicted that
if tears were initially present in these steels they could propagate
6
to failure. For high-cycle fatigue (greater than 1 0 cycles) it was
concluded that failure would always occur at the weld toe of the
cruciform joint for any of the steel.}. This is because of the high
stress concentration at the weld toe.
32
Another study investigated the effects of three different
steelmaking practices on fatigue crack propagation. Heats of ASTM
A 533B steel made by conventional steelmaking practice (CON),
calcium treatment (CaT), and electroslag remelting (ESR) were tested
by use of compact wedge-opening-load fatigue specimens. A ST % RA
of 20% for the CON material indicated that its ST ductility was not
low. The results showed that the fatigue crack growth rate in the
ST direction of the CON material was considerably higher, while the
CaT and ESR materials had significantly Improved isotropy with overall
improvement and reduction in scatter. The improvement in fatigue
crack growth rate was attributed to alteration of inclusion morphology
and distribution. Observations of the fracture surfaces of the CON
30 31
material agreed with the previously reported studies, ' which
showed inclusions affecting the ST direction but not the longitudinal
direction. Tn addition, the same quasi-lamellar tearing decohesion
of inclusions was shown to be responsible for the increased growth
rates in the ST direction.
 40

The following conclusions can be drawn from the results of

the three previously described st-jcles. The fatigue properties of
tear-free steels made by CON can show anisotropy in the ST direction
even though the ST ductility is above levels where lamellar tearing
occurs. However, stress concentrations at the root or toe of welds
would likely be more detrimental in tear-free material unless the ST
ductility is very low. In steels with initially existing lamellar
tearing, the tears appear to be most detrimental to fatigue loading.
6
For high-cycle fatigue (greater than 10 cycles), stress concentrations
at the root or toe of the weld may be more detrimental than existing
tears or poor ST ductility.
17
In ASME C o d e Subsection III-NF, linear supports of classes
1, 2, 3, and MC have fatigue stress range limits (as specified in
Section III Appendices Article XVII-3000) for loading cycles greater
3
than 20 * 10 cycles. The limits are based on the stress category
(type of member and loading) and the number of cycles. The allowable
stress ranges for full and partial penetration T-joints transmitting
a load in the through-thickness direction are penalized by approximately
40 and 60%, respectively. Since these limits account for weld stress
concentrations that can be potentially more detrimental, the design
margin against lamellar tearing during fatigue loading appears
adequate. However, there are no fatigue design limits for plate-
3
and-shell type supports or for loading cycles less than 20 * 1 0 .
In discussions with members of the ASME Code Working Group on Component
Supports (SG-D) (SC III), we were not able to ascertain whether fatigue
limits were omitted in these situations because of their insignificance.

Recent investigation of the effects of dynamic loading with respect

to lamellar tearing has mainly been limited to CVN testing of steels
3
to assess tearing susceptibility. However, some investigations of the
effects of shock loading welded structures have been performed. In a
33
study at the Naval Construction Research Establishment in Scotland,
explosion tests of full-penetration T-joints were performed on
three steels: H.Y. 80, O.T. 35 (a U.K. grade of quenched-and-
tempered C-Mn-Ni-Cr Steel), and B quality (a U.K. grade of fully
 41

killed and normalized C-Mn steel). Both the Q.T. 35 and B quality
steels had histories of lamellar tearing during fabrication, while
the H.Y. 80 is not normally considered susceptible. The results
of the explosion tests showed that the Q.T. 35 and B quality steel
T-joints failed completely by lamellar tearing, while the H.Y. 80
T-joints suffered only limited ductile tearing at the weld toe.
These results demonstrate that dynamic loading can cause the formation
and propagation to failure of« lamellar tears. These tests again show
that if the steel has sufficient ST ductility, the stress concentration
at the weld toe can be the limiting factor in dynamic loading, as was
the situation for fatigue.
5 10 11
Reduced ST CVN properties have been reported ' ' for tear-free
A 515, A 516, and A 533B steels made by CON. In addition, reduced
10
ST dynamic tear (DT) properties have been reported for tear-free
A 533B steel made by CON. These results along with che limited
explosion testing of weldments indicate that the ST directions of
these steels are more susceptible to failure by lamellar tearing and
brittle fracture mechanisms during dynamic loading. Although no
specific investigations of steels with existing tears are known to
this author, the presence of existing cracks can only decrease the
resistance of the material to failure during dynamic leading. Since
selection of material with improved properties is the only design
procedure for ensuring resistance to failure during dynamic loading,
mandatory ST ductility requirements should be specified for Subsection
III-MF components that may oe dynamically loaded.

RECOMMENDATIONS FOR AVOIDING LAMELLAR TEARING DURING

FABRICATION OR SERVICE TRADING

The most effective means for avoiding lamellar tearing is through

the use of material with improved ST ductility. However, selecting
better material increases costs. With the advent of inspection require­
ments and more sensitive UT techniques, the detection of lamellar
tears that result from fabrication has increased. Because these tears
must be eliminated by costly and sometimes ineffective repairs or by
 42

disassembly and replacement of the defective material, the Increased

coses for Improved material are now more justifiable. Improved
materials are already being used In critical service applications.
Requirements for steels to be used In critical components of offshore
structures have already specified lover sulfur contents, minimum ST
tensile strength, minimum ST X RA, and maximum allowable inclusion
21 30 1<>
content. * Table 3 shows some selected material requirements
for critical components in North Sea offshore structures.

Table 3. Selected Material Requirements for Critical

Components in North Sea Offshore Structures

Maximum
Offshore
Application Sulfur ST Z RA
Platform
(2)

B.P. Forties Node tubes 0.010 20 (mean of 6)

15 (min individual)
Shell Brent Nodes 0.015 >35 (mean of 6)
Node type bracing 0.015 >35 (mean of 6)

Figure 15 shows two design paths for avoiding lamellar tearing

during fabrication or service loading. One path is required for
class 1 and the other may be followed for classes 2, 3, and MC
component supports. The information contained in this figure provides
a synopsis of the conclusions from the foregoing assessment of lamellar
tearing. The difference between the two paths is that the one for
class 1 component supports requires the use of material with improved
ST ductility, while the path for the other classes does not. Although
material improvement is preferable, less critical applications, cost,
and availability of material may make the alternate path more desirable
and therefore justify it being an option. However, the reliability
of the path for classes 2, 3, and MC component supports is assured
by two mechanisms. First, the service loading must always be con­
sidered. For dynamic loading, improvement of the material's ST properties
is mandatory. Second, ultrasonic examination of the finished weld
or buttered layer is required.
 43

ORNL-DWC 77-19302
^esigr.
jptiaize -Weld Susceptible Joints First
Fabrication
Seqjecce -Vse s .ibassesblies

Optimize -3alar.ce ,;oint

-Recrient weld
Juict
-Minimize weld size

Consider -Static
Service -Fatigue
Loading -Dynamic • "Improved M a t e r i a l R e q - i r e d

Material
3
S e l e c t i o n ^Required for c l a s s 1 ccr.p^r.er.t s u p p o r t s , -.ptior.ai f o r c l a s s e s <., 3 , MC)

Replace Castings
Hct-Rolled leduce 5 and Z l e v e l s
Steel -Forgir.gs
A l t e r sr.ape, s i z e , t y p e , distribution
Improve Tlrxliisicr.s
-ST I RA
Hot-Polled S t e e l "
Tests -3T CIH ( e s t a b l i s h m e n t
3T P r o p e r t i e s ^Matrix
of Acceptance
Reauce banding limits required)
Improve toughness

Fabrication!
Control
-Do not o/enaatcr. filler metal strength
Filler Metal

Reduce Localized -—Use Balanced Bead Sequence

Contraction Strains

Avoid
Control introduction vf H^
Bobrittleaer.t

Material
Improvement Buttering
Not Selected Technique

z\
Repair

Inspection

"7—Ultrasonic Examination

Fig. 15. Design Paths for Avoiding Lamellar Tearing During Fabrication
of Service Loading of Section III Division 1, Subsection NF Component
Supports.
 44

The following summary highlights the conclusions oresented in

Fig. 15. The avoidance of lamellar tearing must begin at the design
stage. The service loading should be considered, and mandatory ST
material requirements specified for dynamic loading. The joint design
should be optimized by balancing the joint, reorienting the weld if
possible, and minimizing the size of the weld. The fabrication sequence
should be optimized by welding susceptible joints first or by using
subassemblies to minimize the restraint. Material susceptibility to
tearing can be avoided by either replacing the hot-rolled steel with
another product form or selecting material with improved ST properties.
A decision to replace hot-rolled steel with a casting or forging must
weigh the economic consequences. The improvement of the hot-rolled
steel should be aimed at both the inclusions and the matrix. The
addition of rare earth metals or the use of advanced steelmaking
techniques such as calcium treatment or electroslag remelting can be
used to reduce sulphur and oxygen levels and alter the shape, size,
type, and distribution of inclusions. The matrix properties should
be upgraded by reducing the banding and improving the toughness.
The test that is currently most applicable to material assessment
in a production environment is ST Z RA. When this test employs a
sufficient sampling plan with conservative acceptance limits, it
should provide an adequate means of assuring freedom from tearing.
The use of the ST CVN Test provides a method to evaluate the toughness
of the matrix. However, additional work is required to establish
acceptance criteria for this test.

Several considerations are important in reducing the risk of

tearing during fabrication. The filler metal should be controlled
so that it does not overmatch the strength of the base metal. The
localization of contraction strains should be reduced by using a
balanced bead deposition sequence. Embrittlement of the HAZ should
be avoided by using techniques to eliminate the introduction of
hydrogen. The buttering technique can be used to lessen the risk
of tearing in materials without improved ST properties or for repair.
Ultrasonic inspection should be required when material with unimproved
ST properties is used or when the buttering technique is employed.
 45

ASSESSMENT OF ASME CODE REQUIREMENTS

Currently ASME Code Section III Division 1, Subsection NF has

only one requirement to deal with lamellar tearing or poor ST ductility.
This requirement limits the stress at the contact surface of a veld
transmitting a through-thickness load to 50Z of the design stress
intensity (S ) . However, the effect of this requirement has been to
cause designers to increase the size of welds to reduce the stress.
Since increased weld size worsens the risk of lamellar tearing as
previously described, the Code committee is currently planning to
eliminate this stress limitation from Section III Division 1, Subsection
NF. Since the design stress intensity (5^) would still be considerably
i

below the yield strength, as previously discussed, this should not

increase the risk of tearing. As a replacement for this limitation
on stress, a limitation should be placed on the .ize of the weld.
One possible method would be to limit the cross-secti.onal area of a
weld at the contact surface where it produces a tensile load in the
through-thickness direction. The limitation on the weld cross-
sectional ari_a should be based upon the load-carrying area of the
smaller of the two members being joined. An allowance must be made
for additional reinforcement that the welder may add. Typically,
this additional reinforcement could be 10% on either side of the weld.
Therefore, a possible requirement might be that the cross-sectional
area of a weld at the contact: surface where it produces a tensile load
in the through-thickness be less than 1202 of the load-carrying area
of the smaller of the two members being joined.
17
The specifications permitted by ASME C o d e Section III and
Code Case N-71 (formerly 1644-6) for materials that are potentially
susceptible to lamellar tearing are shown in Table 4. It should be
remembered that susceptibility to lamellar tearing depends largely
upon the steelmaking practice used to manufacture hot-rolled plates
and shapes of carbon and low-alloy steel. Materials produced to some
of the specifications listed in Table 4 will be more susceptible
than others. However, this difference in susceptibility is not
necessarily due to different requirements in the spec ifications.
 I- "-•« r«—H*»nl' in-Kin * 1 M

Table A. Material Specifications Permitted by ASME Code Section III-NF

But Potentially Susceptible to Lamellar Tearing

Applicable Component
Support Classes
Specification Description

SA-285 Carbon steel plate for pressure vessels

SA-4A2 Carbon steel plate for pressure vessels
SA-515* Carbon steel plate for pressure vessels
SA-516* Carbon steel plate for pressure vessels
SA-537 Carbon steel plate for pressure vessels
SA-299 Carbon steel plate for pressure vessels
SA-204 Low-alloy steel plate for pressure vessels
SA-302 Low-alloy steel plate for pressure vessels
SA-533 Low-alloy steel plate for pressure vessels
SA-36 6
Structural carbon steel plate and shapes
A 57 2 e
High-strength low-alloy steel plate and shapes
A 588 High-strength low-alloy steel plate and shapes
SA-612 Carbon steel plate for pressure vessels
SA-283 e
Structural carbon steel plates

From ASME Boiler and Pressure Vessel Code Sect. Ill, Div. I Appendices,
Tables 1-1.1 and 1-11.1; ASME Code Case N-71, Table 1.
From ASMF Boiler and Pressure Vessel Code Sect. Ill, Div. 1 Appendices,
Tables 1-7.1 and 1-12.1; ASME Code Case N-71, Table 2.
c
From ASME Boiler and Pressure Vessel Code Sect. Ill, Div. I Appendices, Table 1-8.1.
From ASME Boiler and Pressure Vessel Code Sect. Ill, Div. I Appendices, Table 1-10.0.
g
History of lamellar tearing, reported by S. Ganesh and R. D. Stout, Weld. J. (Mi-am)
•35(11): 341-s-354-s (November 1976), or D. A. Canonlco, ORNL, private communication.
 47

These specifications generally allow a wide range of steelmaking

processes (open-hearth, basic-oxygen, electric-furnace, vacuum-arc
reaelt, or electroslag remelt), deoxidation practices (rimmed, capped,
semikilled, killed, killed plus aluminum for fine grain practice),
sulfur content (0.04-0.05Z maximum), and silicon content (0.15-0.30Z).
Therefore disallowing the use of certain specifications will not be
an adequate solution to the lamellar tearing problem. However, the
use of an ST Z RA requirement will force the designer and the steel
producer to select the steelmaking practices necessary to improve
ST ductility. Table 4 shows that five of the material specifications
allowed for class 1 component supports have a reported history of
lamellar tearing. Because of the critical nature of class 1 components,
this situation highlights the need for mandatory ST ductility requirements
in class 1 component support materials that will contain c weld
transmitting a load in the through-thickness direction.

Since the redesign of weld joints, other than completely eliminating

through-thickness strain in rolled plates or shapes, cannot always
eliminate tearing in susceptible material, mandatory joint designs
do not seem appropriate. However, the addition of pre"erred joint
17
designs to C o d e Subsection III-NF Fig. NF-3291(a)-l requirements
17
for plate and shell supports and the C O D E Section III Appendix XVII-2450
requirements for linear supports would be appropriate. Examples of some
preferred joint designs are discussed in a previous section.
The ASME Code Subgroup on Fabrication and Examination (SC III) is
currently formulating requirements for welds that transmit a through-
thickness load in rolled plates and shapes of Subsection III-NF
component support primary members greater than 16 mm (5/8 in.) in
3
thickness. The currently 'proposed requirements provide options to
either ultrasonically examine finished welds, weld on an ultrasonically
examined inlay or overlay, or test the base metal to meet an ST % RA
of 20Z for a single test. In addition, the minimum specified tensile
strength of the filler metal is not allowed to exceed that, of the
base metal by more than 83 MPa (12 ksi). These requirements are
aimed at the Subsection III-Nf fabrication and installation article
 48

and do not interface with the materials and design articles as

recommended in Fig. 15. The proposed requirements should Incorporate
»
design considerations of component support class and service loading.
Mandatory materials requirements should be Included for class 1
component supports or service conditions that may experience dynamic
loading.
The decision to limit requirements to material greater than 16 mm
1
(5/8 in.) in thickness is not totally supported by a reported survey
of fabricators. In this survey lamellar tearing was reported by 20Z
of the fabricators in situations in which the plate thickness did
not exceed 13 mm (1/2 in.). Although 95Z of the fabricators reported
tearing to occur in thicknesses between 13 and 64 mm (1/2 and 2 1/2 in.),
the potential for tearing in highly restrained thin plates cannot be
3
overlooked. In addition, a recent study ' has shown that the inclusions
in thin plate can be more elongated than those in heavy plate if the
rolling ratio (ratio of primary rolling to cross rolling) is not
kept low. As a result of the above considerations, it may be prudent
to require Improved ST ductility in all thicknesses of class 1 component
supports, and require the use of one of the three options in thicknesses
above 16 mm (5/8 in.) for all other classes.
As previously discussed, the optional use of ultrasonic examination
of finished welds or welding on an ultrasonically examined buttered
layer for materials that do not have improved ST ductility is an
acceptable procedure if component support class and service loading
are considered. The methods described for ultrasonic examination and
buttering in the proposed requirements appear to follow the procedures
recommended in the literature. In addition, the limitation on filler
metal strength is a good requirement.
The proposed requirements for specifying a minimum ST Z RA need
to be revised. As previously recommended, the ST X RA requirements
should be limited to steels with yield strengths less than 400 MPa
(58 ksi). The requirement for a single tensile test with a minimum
ST X RA of 20Z needs to be modified to provide a more conservative
acceptance limit. The need for a higher ST X RA can be justified
 49

because of the large variations that commonly occur between 0 and

37
20Z. Dolby has recently reported that although Welding Institute
experience has shown that an ST Z RA of 20Z (for a set of six speciaens)
is adequate to avoid tearing in highly restrained joints, a aore
conservative ainiaua ST Z RA such as 2SZ is necessary to make allowance
for such factors as inclusion segregation. In addition, a recent
3
Welding Research Council Bulletin, ' which presents an international
bibliography on through-thickness properties and laaellar tearing,
points out the need for statistical control over ST Z RA requirements
because of their variability. As an example, recent changes in
specifications for materials for offshore construction are cited.
Whereas two to three years ago an ST Z RA of 2QZ was used, recent
specifications require a minimum individual value of 10Z with a
35Z minimum average for a set of six specimens. ' 3
A minimum ST Z RA
of 25Z is the acceptance limit used for the most reliable steel
21
grades in French and German standards. The testing of a set of
six specimens should also be required to improve the statistical
representativeness of the test. Existing specifications are currently
requiring sets of 3 to 6 specimens. The size of specimens and the
extraction procedure should be standardized. Since it is advantngeous
to use a large specimen diameter and possibly include the plate
surface regions in the specimen gage length, these factors should
be considered when establishing standard specimen size and extraction
procedures. In addition, requirements should be established for the
welding of grip attachments to plate surfaces. The range of product
thicknesses from a given heat that a set of tests can represent must
36
also be specified. As previously reported, the amount of reduction
can affect the degree of inclusion elongation and thereby influence
ST ductility.

CONCLUSIONS

A need is evident to establish requirements to reduce the risk

of lamellar tearing in critical applications such P.S N-iclear Power

!
 so

Plants. However, because of the controversy surrounding the variou_

aspects of laaellar tearing, it has been difficult to establish
universally accepted requireaents. The design paths shown in
Fig. IS along with the specific recoaaended acceptance Halts and
testing conditions provide usable requireaents that are based upon
both fabrication experience and research prograas. While it aay
appear to soae that these requireaents are too conservative, this
apparent ccnservatisa is needed to bridge the gaps in the controversal
r^pects of laaellar tearing.
The results of this assessaent indicate that the iaproveaent
of material properties is undoubtedly the best approach for reducing
the risk of laaellar tearing. However, it was also recognized that
under certain service loading conditions in noncritical applications,
the use of ultrasonic examination or buttering can provide acceptable
methods for reducing the risk of tearing. This situation highlights
the fact that the requireaents that are recoaaended as a result of this
assessaent were aiaed at both reducing the risk of laaellar tearing and
providing flexibility to fabricators when possible.

ACKNOWLEDGMENT

1 would like to acknowledge D. A. Canonico and W. A. Simpson, Jr.

for reviewing this paper, S. Peterson for editing, and Gail Golliher
for typing the manuscript.

REFERENCES

1. J.CM. Farrar and R. E. Dolby, "Laaellar Tearing in Welded Steel

Fabrication," in Phase I of Investigations into Lamellir Tearing.
A Compendium of Reports from a Sponsored Research Prograrme, The
Welding Institute, Cambridge, England, March 1975.
2. S. Ganesh and R. D. Stout, "Material Variables Affeccing Lamellar
Tearing Susceptibility in Steels," Weld. J. (Miami) 55(11):
341-8-354-s (November 1976).
3. J.E.M. Jubb, "Lamellar Tearing," W:l :. Res. ?owt?. Bull. 168:
1-14 (December 1971).
4. "Commentary on Highly Restrained Welded Connections." A1SC FKJ.
1973: 61-73.
5. A. Hattangadi and B. Seth, lar&llar Tezrir? ir Fillet WeZ zr.eKis c~
Pressure Vessel Fzbri^itions, American Society for Metals Technical
Report System Ho. 76-50 (1976).
6. J.CM. Farrar, R. £. Dolby, and R. C. Baker, "Lamellar Tearing in
Welded Structural Steels," Weli. C. tXerj Y?rk) 48:7 274-s-282-s
(July 1969).
7. B. Kato and K. Morita, "On the Static Anisocropy of the Structural
Steel Plate," pp. 66—69 in The Fourteenth Jspan Cozgress on
yjitevlals Research - Xetalli? llateriils, March 1971.
8. J.CM. Farrar and R. E. Dolby, "Lamellar Tearing in Welded Steel
Fabrication — The Role of Sulfide Inclusions," pp. 252-68 in
Sulfite Inclusions in Steel, American Society for Metals, Metals
Park, Ohio, 1975.
9. S. Kanayawa et al., "On the Assessment of the Lamellar Tearing
Susceptibility of Steel Plate," TTIKS. J:;n. Weli. Soc. 5(1): 2-14
(April 1974).
10. A. D. Wilson, "The Effect of Advanced Steelmaking Techniques on
the Inclusions and Mechanical Properties of Plate Steels," presented
at Syripos"*4n on Toiikness characterization ini Specificitions for
HSIA and Structural Steels, AIME Annual Meeting March 6-10, 1977.
11. R. P. Oates and R. D. Stout, "A Quantitative Weldability Test for
Susceptibility to Lamellar Tearin ," W^d. ,\ (y.innii
R 52(11):
481-s-491-s (November 1973).
12. J.CM. Farrar, ''Inclusions and Susceptibility to Lamellar Tearing
of Welded Structural Steels," Veld. J. {Xivri* 53(8): 321-s-331-s
(August 1974).
13. S. Ganesh and R. D. Stout, "Effect of Welding Variables on Lamellar
Tearing Susceptibility in the Lehigh Test," Weli. .1. (V.iv*.i) 56(3):
78-S-87-S (March 1977).
 52

14. Private communication with T. G. Davey, The Welding Institute,

April 7, 1977,
15. J.CM. Farrar, E. N. Gregory, and R. E. Dolby, "The Problem of
Laaellar Tearing. The Fabrication of Welded Testpieces: Mechanical
Test Criteria for Avoiding Lamellar Tearing and the Effects of
Soae Welding Procedural Variables on Lamellar Tearing," in
Phase VI of Investigations into Lamellar Tearing. A Compendium
of Reports from a Sponsored Research Programme, The Welding
Institute, Cambridge, England, March 1975.
16. Welding Handbook. Section One, Sixth Edition, AWS 5.40-5.43 (1968).
17. ASME Boiler and Pressure Vessel Code, 1977 ed. with Addenda
through Summer 1977, American Society of Mechanical Engineers,
New York, 1977.
18. J.CM. Farrir et al., "The Assessment of Susceptibility to
Lamellar Tearing by Ultrasonic Means," and a series of six progress
reports "The Problem of Lamellar Tearing. The Assessment of
Susceptibility to Lamellar Tearing by Ultrasonic Means," in
Phase V of Investigations into Lamellar Tearing. A Compendium
of Reports from a Sponsored Research Programme, The Welding Institute,
Cambridge, England, March 1975.
19. J.CM. Farrar, "The Ultrasonic Assessment of Susceptibility to
Lamellar Tearing - The State of the Art," Weld. Inst. Res. Bull.
17(8): 205-08 (August 1976).
20. C. Tamburelli, "Use of Ultrasound in Assessing the Susceptibility
of Steel to Lamellar Tearing," Non-Destr. Test. (Guildford, En^l.)
10(1): 3-8 (February 1977), as cited by Current Awarenees Bull.
8(13): 3 (July 8, 1977), Metals and Ceramics Information Center,
Battelle, Columbus, Ohio.
21. Private communication with T. G. Davey, The Welding Institute,
Aug. 11, 1977.
22. Notes from the Meeting of Sponsor Group 5513, The Significarice
of Lamellar Tearing, The Welding Institute, June 1, 1977,
 53

23. J.CM. Farrar, "The Problem of Lamellar Tearing. The Relationship

Between Short Transferse % RA Values and the Incidence of Lamellar
Tearing," in Phase IV of Investigations into Lamellar Tearing.
A Compendium of Reports from a Sponsored Research Programme,
The Welding Institute, Cambridge, England, March 1975.
24. B. J. Ginn and J.CM. Farrar, "The Problem of Lamellar Tearing.
The Development of a Slice Bend Test to Assess Susceptibility to
Lamellar Tearing," in Phase IV of investigations into lamellar
Tearing. A Compendium of Reports from a Sponsored Reserach
Programme, The Welding Insittute, Cambridge, England, March 1975.
25. M. L. Drury and J.E.M. Jubb, "Lamellar Tearing and the Slice Bend
Test," Veld. J. (Miami) 52(2): 88-S-95-S (February 1973).
26. J.CM. Farrar and R. E. Dolby, "An Investigation into Lamellar
Tearing," Met. Constr. Br. Weld. -J. 1: 32 (1969).
27. J.CM. Farrar, "The Problem of Lamellar Tearing. The Assessment
of Susceptibility to Lamellar Tearing Using a Short Transverse
Notched Bend Test," in Phase IV of Investigations into Lamellar
Tearing. A Compendium of Reports from a Sponsored Research
Programme, The Welding Institute, Cambridge, England, March 1975.
28. D. N. Elliott, "Lamellar Tearing in Multi-Pass Fillet Joints,"
Veld. J. (New York) 48(9): 409-s-416-s (September 1969).
29. T. G. Davey et al.. The Significance of Lamellar Tearing, Progress
Report for Group-Sponsored Project 5513, The Welding Institute,
May 1977.
30. C. J. Adams and E. P. Popov, "Through Thickness Fatigue Properties
of Steel Plate," Weld. Res. COUKC. Bull. 217: 1-13 (July 1976).
31. S. J. Maddox, "The Problem of Lamellar Tearing. The Significance
of Lamellar Tears with Respect to Fatigue Loading," Phase VII of
Investigations into Lamellar Tearing. A Compendium of Reports
from a Sponsored Research Programme, The Welding Institute,
Cambridge, England, (March 1975).
32. A. D. Wilson, "Fatigue Crack Propagation in A533B Steels," Paper
76-WA1PVP-6 (1976), presented at the Winter Annual Meeting,
New York, Dec. 5—10, 1976, of the American Society of Mechanical
Engineers.
 54

33. A.D.E. Thoason, P. R. Christopher, and J. Bird, "Short Transverse

Properties of Certain High-Strength Steels," J. Eng. Ind. 90:
627-35 (1968).
34. J. A. Chapaan, A. Clark, and P. R. Rirkwood, "Steels for North
Sea Structures," presented at Welding in Offshore Structures,
Newcastle, UK, 1974.
35. Proposed paragraph NF-4430 accepted at the Nov. 1, 1977, meeting
of the ASME Code Subgroup on Fabrication and Examination (SC III).
36. A. D. Wilson, "The Influence of Thickness and Rolling Ratio on
the Inclusion Behavior in Plate Steels," presented at the ASM
1977 Materials Conference, Chicago, 111., Oct. 25-27, 1977.
37. R. E. Dolby, "The Weldability of Low Carbon Structural Steels,"
Weld. Inst. Res. Bull. 18(8): 209-16 (August 1977).
38. D. H. Skinner and M. Toyama, "Through Thickness Properties and
Lamellar Tearing (a Bibliography)," Veld. Res. Counc. Bull.
232: (Hoveraber 1977).