Optimizing Repair Welding Techniques

in Cast Steels—Part II

Hz-induced cracking can be prevented by using low

hydrogen practice if the maximum HAZ hardness is below

Re 35, and all steel castings should be normalized before repair welding

BY D. K. AIDUN AND W. F. SAVAGE

ABSTRACT. Four different heats of cast 1. A metallurgical requisite. For crack- induced residual stresses in all but the
steel were used for hydrogen-induced ing to occur, a crack-sensitive microstruc- simplest of weldment geometries are
cracking studies using low- and high- ture must be present, and, in general, the usually sufficient in magnitude to initiate
hydrogen practice in simulated repair susceptibility to hydrogen-induced crack- cracking if the other requisites are met.
welds. It was shown that to prevent ing increases with increase in the hard- Furthermore, unless stress relief heat
hydrogen-induced cracking where good ness of the microstructure. treatment is performed after welding, the
low-hydrogen practice is not possible, 2. A chemical requisite. For cracking to addition of service usually raises the local-
the maximum "safe" level of hardness in occur, the level of diffusible hydrogen ized stresses in the weldment above the
the heat-affected zone should be below present in the crack-susceptible micro- critical level for crack propagation.
Re 35. structure must exceed a critical concen- Because most welded structures expe-
Metallographic examination of these tration, which decreases sharply as the rience service in the crack-sensitive tem-
cast-steel weldments showed seven dis- yield strength of the microstructure perature range, this requisite is usually
tinct regions from the fusion zone out present increases (Ref. 1). met. However, it is possible to take
toward and including the unaffected base 3. A mechanical requisite. Although advantage of this requirement in t w o
metal. These regions are: cracking has been shown to initiate at ways:
1. Fusion zone. lower stress levels, it appears that crack 1. Postweld soaking. It has been
2. Unmixed zone. propagation requires localized stresses of shown by Adams (Ref. 4) and others that
3. Coarse-grained homogeneous HAZ. approximately yield-strength magnitude an immediate postweld soak for 30
4. Fine-grained homogeneous HAZ. (Ref. 2). minutes (min) to one hour (h) at a temper-
5. Partially transformed region. 4. An environment requisite. Hydro- ature of above 390°F (200°C) will usually
6. Spheroidized region. gen-induced cracking has been shown to prevent hydrogen-induced cracking. This
7. Unaffected base metal. result from the extraordinary reduction in is attributed mainly to the reduction in the
All as-cast heats with carbon contents the notched-tensile strength of steels level of diffusible hydrogen made possi-
ranging from 0.21 to 0.32 wt-%, upon caused by the presence of a relatively ble by the greatly increased diffusivity of
exposure to a 5-second arc time, exhib- small amount of diffusible hydrogen (Ref. hydrogen at 390°F (200°C). Thus, by the
ited maximum hardness levels of about 3). Since this loss in strength in the pres- time the weldment is allowed to return to
70 Re in the networks of martensite ence of a notch occurs only in the range the crack-sensitive temperature range,
surrounding the islands of untransformed - 1 4 8 to 390°F ( - 1 0 0 to 200°C), hydro- the residual level of diffusible hydrogen is
ferrite in the partially transformed region. gen-induced cracking requires that the rendered below the critical level for crack
This was attributed to the inhomogeneity weldments be exposed to temperatures initiation.
of the as-cast structure. However, for within this range in the presence of the 2. Cryogenic storage. In conducting
low-carbon cast steels this problem was first three requisites. laboratory studies, it is convenient to be
eliminated by normalizing the alloy Although in theory hydrogen-induced able to store welded specimens under
before the welding operation. For cast cracking can be prevented by elimination conditions which preclude either loss of
steels of about 0.30 wt-%, in addition to of any one of the above requisites, in hydrogen or the formation of hydrogen-
homogenization, a postweld heat treat- practice it is usually possible to exert induced cracks until controlled experi-
ment would be advisable. control over only the first t w o . This ments can be initiated. This has been
results from the fact that the thermally- shown (Ref. 5) to be possible by transfer
of the weldments to liquid nitrogen
Introduction
(-320°F) (-195.8°C boiling point) as
D. K. AIDUN is Assistant Professor, Mechanical
It is generally recognized that four soon as they reach ambient temperature
and Industrial Fngineering Department, Clark-
requisites must be met simultaneously for son University, Potsdam, New York; and W. F. and storing under liquid nitrogen until the
hydrogen-induced cracking to occur. experiments begin.
SA VA GE is Professor Emeritus, Rensselaer Poly-
These can be summarized as follows: technic Institute, Troy, New York. The critical hardness level is influenced

WELDING RESEARCH SUPPLEMENT 197-s

 (260 X 108 X 19 mm). From these cou-
Table 1—Chemical Composition, Initial Hardness (IH), and Ideal Critical Diameter (D[) of Heat pons, specimens of 10 X 2 X V2 in.
32, 27, 1, and 0 (254 X 50.8 X 13 mm) were machined
for the hydrogen-induced cracking test.
Composition, wt-% The surface on which the welds were to
32 27 1 0 be deposited was then polished with 240
C 0.32 0.31 0.21 0.26 grit SiC paper and degreased with ace-
Mn 1.15 1.57 1.24 0.74 tone.
P 0.017 0.013 0.018 0.021 Polishing and degreasing are critical
S 0.017 0.013 0.031 0.028 and important for the hydrogen-induced
Si 0.54 0.50 0.51 0.40 cracking test when using low-hydrogen
Ni 0.03 0.07 0.04 welding practice. These cleaning opera-
Cr 0.13 0.14 0.20 0.31 tions mitigate hydrogen absorption in the
Mo 0.22 0.10 0.03 0.02 weld by removing hydrogenous materials
Cu 0.04 0.04 0.04 0.03
Al 0.078 0.054 0.11 from the surface of the specimen.
0.075
IH, Re 51 51 46 48
D|, in. 3.17 3.51 2.05 1.61 Procedure
The electrode type, hydrogen content,
heat treatments, and the heats of steel
by the toughness and ductility of the cial surface imperfections. The rapid used in this part of the investigation are
microstructure of the weld heat-affected cooling rates, which are associated with summarized in Table 2. The level of
zone (HAZ). O f the various possible such "cosmetic" repairs, coupled with diffusible hydrogen in welds was mea-
microconstituents, untempered martens- inadequate control of sources of hydro- sured by the RPI silicone-oil extraction
ite is the most brittle and, therefore, the gen such as improper handling and stor- technique (Ref. 6). Two levels of diffus-
most susceptible to hydrogen-induced age of consumables, greatly increase the ible hydrogen were studied:
cracking. The other microconstituents, probability of hydrogen-induced crack- 1. Low-hydrogen practice. After re-
listed in order of decreasing susceptibility, ing. moval from hermetically sealed shipping
include lower bainite, upper bainite, Underbead cracking is one of the most containers, the electrodes were trans-
pearlite, spheroidite, and ferrite. insidious forms of hydrogen-induced ferred immediately to a well-ventilated
In general, the toughness and ductility cracking. This is because it often is lo- holding oven and maintained at 350°F
both of individual and particular combi- cated entirely beneath the weld and fails (177°C) until just before use. This proce-
nations of microconstituents is decreased to emerge at the surface where it can be dure gave an average level diffusible
by increasing the carbon content of the detected by nondestructive inspection. hydrogen of approximately 3 ppm (3.3
prior austenite. For this reason, the sus- Unfortunately, the geometry of most cc/100 g of weld metal).
ceptibility to hydrogen-induced cracking repair welds is such as to make this a 2. High-hydrogen practice. The elec-
is directly related to the carbon content dominant form of hydrogen-induced trodes were removed from the open
of the steel. For a given carbon content, cracking in steel castings. If undetected, shipping container, dipped in water for
increasing the percentage of martensite such cracks may experience slow crack 30 s and used immediately. This practice
in the weld heat-affected zone lowers growth in service until the critical crack gave an average diffusible hydrogen level
the critical hydrogen concentration size is reached and catastrophic failure of 30 ppm (33.3 cc/100 g of weld met-
required to cause hydrogen-induced occurs. al).
cracking. The 10 X 2 X l/ 2 in. (254 X 50.8 X 13
Carbon is the least expensive of all mm) bars were welded using various arc
Materials and Specimen
alloying elements. Because of this, steel times and weld distances with both low
Preparation
foundries tend to increase the carbon and high-hydrogen practice. After cool-
content of their steels in order to achieve Four heats of cast steel were used in ing to ambient temperature, the weld-
higher strength. The increase in carbon this investigation. Table I lists the chemical ments were immediately loaded in the
content increases the ideal critical diame- composition, initial hardness, IH (the hard- varestraint apparatus (Ref. 7) and a 2%
ter, D|, more than any other alloying ness of martensite), and ideal critical augmented strain was then applied.
element; in addition, it also increases the diameter, D, (based upon 50% martens- During this process, an acoustic-emis-
hardness of any martensite formed. ite), for all four heats. sion transducer was mounted on the
A few years ago, the railroad industry The four heats, supplied in the form of specimen surface adjacent to the weld
began to investigate the failure of cast cast coupons, had dimensions that were bead, as shown schematically in Fig. 1, to
components such as couplers, yokes, and approximately 10Vi X4V4 X 3 /i in. monitor crack initiation and propagation.
knuckles. As a result of this investigation,
it was concluded that hydrogen-induced
cracking was a contributing cause of
Table 2—Summary for Electrodes, Hydrogen Content, Heat Treatments, and Steel Heats Used
failure. It seemed likely that hydrogen-
in Investigation
induced cracking could result from
improper welding procedures used in
Diffusible H2 in welds made with
repairing casting defects in these struc-
3/16 in. (4.76) diam. electrodes:
tures. It was suspected that, in the Moist E7018(a> 30 ppm
absence of suitable control of repair- Baked E7018(b) 3 ppm
welding procedures, crack susceptible Heats 32, 27, 1, 0 As-received
heat-affected-zone microstructures could Heats 32, 27, 1,0 Normalized at 1750°F (954°C)
easily be produced. Heat 27 Normalized and postweld heat treatment
Unfortunately, repair welds in steel at 11003F(593°C)
castings are often performed solely for '"'Electrode dipped in water for 30 seconds.
"cosmetic" reasons to eliminate superfi- (b)
Electrode stored continuously at 350-F (177 C C) after removal f r o m sealed container.

98-sl APRIL 1985

 Because of the high ductile-brittle transi-
tion temperature (DBTT) of these heats,
- < « • •

the hydrogen-induced cracking could not

be performed at liquid-nitrogen tempera-
X
ture, - 3 2 0 ° F (-196°C). :A !#•
fjk
Specimens were then cut and
mounted for metallographic examination
and microhardness measurements.

Hydrogen-Induced Cracking Tests

Equations described in an earlier paper 1 ftot
(Part I to this paper — Ref. 8) were used to
select welding conditions which would
""31
®
give maximum weld heat-affected zone
hardness levels ranging from about 25 to Fig. 2 — Typical underbead crack in the coarse-
50 Re in the coarse-grained region at the grained region of the HAZ of a stationary weld
fusion boundary. Welds were made in made on heat 27 using high-hydrogen prac-
heats 32, 27, 1 and 0 using the selected tice. Arc current - 220 A; arc voltage-25 V;
welding procedures to make simulated arc time-5 s; section thickness- 'A in. (12.7
mm); cooling rate at 900°F (482°C)-49°F/s
repair welds with both high- and low-
(27.2°C/s); hardness-52 Re (50 Re pre-
hydrogen practice. dicted). 2% nital, X500 (reduced 46%, on
The welded specimens were loaded in reproduction)
the varestraint test apparatus. Then, after
attaching the acoustic-emission transduc-
er, they were subjected to a 2% aug- For welds made using high-hydrogen
mented strain for 2 h in order to simulate practice the maximum "safe" hardness
the mechanical factor by creating yield- level was found to be about 33 Re.
strength stresses in the outer fibers of the Welds made with conditions that
specimen. produce hardness levels of less than 35 Fig. 3 — Effect of cooling rate on microstructure
of repair weld HAZ in heat 1. A -stationary
The acoustic-emission data provided Re did not exhibit detectable cracks in the
weld (stationary arc); B — linear weld. Welding
qualitative information on the initiation coarse-grained region of the heat-af- conditions: for A and B — 220 A arc current, 25
and propagation of hydrogen-induced fected zone, even with 30 ppm of diffus- V arc voltage, V2 in. section thickness; for A
cracking. However, quantitative correla- ible hydrogen present. only-5 s arc time, cooling rate of 39°F/s @
tion between the acoustic-emission With low-hydrogen practice, no crack- 900"F, 51 Re max. hardness; for B only- 10
counts and the severity of cracking was ing was detected in the coarse-grained ipm travel speed, 2.0 in. weld length, 18°F/s
not found. This was believed at the time cooling rate @ 900"F, 31 Re max. hardness.
region until hardness levels of approxi-
to result from extraneous noise in and 2% nital etch, X500 (reduced 35% on repro-
mately 45 Re were produced. duction)
about the laboratory that caused errone- Figure 2 shows a typical underbead
ous counts. However, for reasons that crack, which initiated in the coarse-
are discussed in a later section, this may grained region of the weld heat-affected Figure 3 shows weld heat-affected-
be only partially responsible for the lack zone. The microstructure around the zone microstructures which experienced
of correlation. crack is predominantly martensitic. maximum cooling rates of 18 and 39°F/s
(10 and 21.7°C/s), respectively, at 900°F
(482 °C). The structure in Fig. 3A consists
To Acoustic of a mixture of martensite (identified by
Weld Bead
-» Emission the letter M), lower bainite (identified by
-T^P—i Recorder the letter BL) and upper bainite (identified
by the letter Bu).
The measured maximum hardness in
the martensitic regions of the specimen in
Fig. 3A was Re 51, whereas the calculated
value of IH for this material (heat 1) is only
45 Re. This means that the carbon con-
tent of the prior austenite was probably
inhomogeneous and was higher than the
nominal carbon content in some regions,
thus explaining the higher than expected
maximum hardness. This is entirely possi-
aug 2 R ble because the short time-at-tempera-
ture could prevent complete homogeni-
zation from occurring even at tempera-
tures just below the solidus. The structure
in Fig. 3B consists of upper and lower
bainites with maximum hardness level of
Re 31.
SIDE VIEW The major difference between cast
and wrought steel products lies in the
Fig. 1 - Schematic diagram of bending section of the varestraint apparatus and location of acoustic nature and severity of the compositional
emission transducer with respect to weld heterogeneity. In both types of product,

WELDING RESEARCH SUPPLEMENT 199-s

 .*--'

'
-#>

*
'«*•
4 £ C 0 E F
Fig. 5-Region A in Fig. 4 at X500 (reduced
fig. 4 — Microstructure of 5 s stationary repair weld in heat 27 with KHN (50 g load) hardnesses as 46% on reproduction). 2% nital etch
follows: A- 168; 6-280; C-403; D-403 prior pearlite, 197 prior ferrite; £-615 prior pearlite, 84
prior ferrite; F— 163 prior pearlite; C-(not stated). Nital etch, X50 (reduced 49% on reproduc-
tion)

the initial solidification process is accom- weldment shows seven distinct types of
panied by dendritic segregation with the microstructures labelled A through C in
dendrite interstices being enriched in Fig. 4:
those elements which depress the liquid- A — fusion zone. Figure 5 shows the
us temperature of the alloy. However, microstructure of this region as revealed
during soaking at high temperature in by a 2% nital etch at X500. The micro-
preparation for hot-working, partial structure consists of Widmanstatten fer-
homogenization occurs, and the pattern rite and finely laminated pearlite. The
of the residual segregation is altered by hardness of this low-carbon weld metal
the hot-working operations. Thus, in the was only 168 KHN.
case of hot-rolled plate products, the B — unmixed zone (Ref. 9). The top Fig. 6 - Region B in Fig. 4 at X500 (reduced
combination of thermal and mechanical section of Fig. 6 shows this zone at X500. 54% on reproduction). 2% nital etch
treatment forms alternating bands of sol- Because mechanical mixing is incomplete
ute-rich and solute-lean material parallel in this zone, the substitutional alloy con- mottled appearance is indicative of inho-
to the rolled surface. tent approximates that of the base metal. mogeneity. Figure 8 shows a portion of
In the case of steel castings, the pattern Carbon, on the other hand, diffuses so this region at X500. The light gray regions
of segregation is controlled by the den- rapidly that the carbon content is inter- are fully martensitic and exhibited a hard-
dritic-arm spacing, which increases as the mediate between that of the weld (0.07 ness of 403 KHN. The intermediate
section thickness increases. Since no wt-%) and that of the base metal (0.31 regions which consist of a mixture of
mechanical working is involved, homoge- wt-%). The region consists of bainite, elevated-temperature transformation
nization of castings can only be achieved Widmanstatten ferrite and very fine products exhibited a hardness of 197
by diffusion-controlled thermal treat- pearlite structures with measured hard- KHN.
ments. When the dendrite-arm spacings ness of only 280 KHN. Apparently region D experienced peak
are large, the increased diffusion dis- C—homogenized coarsed-grain re- temperatures above the effective A3 long
tances require longer times and higher gion. Figure 7 shows the microstructure enough to complete the transformation
temperatures to achieve a reasonable of this region at X500. The uniform to austenite but for too short a time to
degree of homogenization. Thus, it is response to the etchant within this region experience homogenization of even the
customary to normalize cast steels prior in Fig. 4 indicates it to be reasonably interstitial carbon. The austenite grain size
to the final heat-treatment operations homogeneous as a result of the high peak in this region ranges from slightly coars-
which are designed to control the temperatures experienced in this region. ened to fully refined and the hardenabili-
mechanical properties of the castings. The microstructure is fully martensitic and ty varies accordingly.
Figure 4 shows the microstructural exhibited a hardness of 403 KHN. E—partially transformed region (PTR).
changes from the fusion zone at the left D — inhomogeneous fully transformed This region is characterized by regions of
outward to the unaffected base metal at region. Within this region the response to nearly eutectoid carbon which corre-
the right. The panoramic view of the the etchant in Fig. 4 is uneven and the spond to the original networks of pearlite

Fig. 7—Region C in Fig. 4 at X500 (reduced Fig. 8 —Region D in Fig. 4 at X500 (reduced Fig. 9-Region E in Fig. 4 at X500 (reduced
52% on reproduction) 54% on reproduction). 2% nital etch 53% on reproduction). 2% nital etch

100-s|APRIL 1985
 *f' > • ^ *.,*
- - .»r»-«
4 -. v *^
•E
— m *& •
" " "' ...IIP TR* -
• - - % ^

' :,'A
•r tf

%* ' ' 'tjf »"*P"> •" "

jr ' *! •Ai:'.:~3A
-. S * (
.»: :,.**•£ - £• Fig. 14—Microstructure of 5 s stationary repair weld in heat 0. Nital etch, X75 (reduced 42% on
% . 10-Region F in Fig. 4 at X500 (reduced reproduction)
46% on reproduction). 2%, nital etch
untransformed ferrite islands. Figure 9 11 shows the microstructure of this zone
shows the location of E at X500. at X500. Note that the pearlite lamina-
• • •

F—spheroidized region. In this region, tions can be resolved in some areas at

the peak temperature experienced this magnification.
• between the effective A i and about Figure 12 summarizes the results of a
***• ..J^.- 1 * -. 900-1000°F (482-538°C). The lamellar microhardness traverse across the region
*i*"' * " "
j-' ' '-' carbides within the pearlite nodules shown in Fig. 4. Attention is drawn to the
spheroidize while the original ferrite large spread between the maximum and
grains remain unchanged. Figure 10 minimum hardness values observed in
• M ' ^ | j shows this microstructure at X500. The the partially-transformed region. This
average hardness is 163 KHN in the prior reflects the nonuniform carbon content
'• 40». pearlite and about 84 KHN in the prior of the martensite formed in this region.
ferrite. Note that the areas of untransformed
C—unaffected base metal. This micro- ferrite in this region exhibited hardness
f/g. 11-Region G in Fig. 4 at X500 (reduced
structure consists of continuous networks values ranging from 77 to 94 KHN. An
52% on reproduction). 2% nital etch
of pearlite nodules at the dendrite inter- approximate scale of Re hardness is
in the as-cast microstructure. This region stices as a result of the segregation of included at the right side of Fig. 12 for
transforms to high-carbon austenite carbon and alloying elements to the last comparison with calculated hardness
whenever the peak temperature exceeds material to solidify. This means, in effect, data presented earlier.
the effective A-i, and the short times- that the as-cast base metal, which has a
at-temperature prevent significant redis- nominal carbon content of 0.31%, be- Homogenization Studies
tribution of the carbon. Therefore, the haves like a composite material. As a
continuous prior-pearlite networks form composite material, it consists of islands The results summarized in the previous
continuous networks of high-carbon aus- of low carbon ferrite grains (~0.025% C) section clearly indicate the dangers of
tenite; upon cooling, these have sufficient completely surrounded by nearly contin- repair welding as-cast steel structures. It is
hardenability to transform to continuous uous networks of material with carbon
networks of high-carbon martensite. The contents ranging from approximately
hardness of 615 KHN is in the high- 0.5% (the carbon content of liquid steel at
carbon martensitic regions and the low the peritectic temperature) to nearly
values of 84 KHN are encountered in the eutectoid (~0.8%) composition. Figure

:,(,'
UNTRANSFORMED
REGION OF
AS CAST STRUCTURE
<
o
_i

Q
Q IOOO CC
<
I

Q
CC
<
I
TYPICAL HARDNESS OF
o PEARLITE NODULES
o
FfRJITS :".»___!_

_L_
0 0.050 0.I00 0.I50
Fig. 13 — As-cast and normalized heat 0 micro-
DISTANCE FROM FUSION BOUNDARY-inches structures: A — as-cast; B —normalized. 2% nital
Fig. 12 -Results of hardness traverse across regions shown in Fig. 4 etch, X250 (reduced 44% on reproduction)

WELDING RESEARCH SUPPLEMENT 1101-S

 IOOO '*

o
o

o
o
m
9 500-
Fig. 18 — Tempered martensite in Fig. 17 at
cn X500 (reduced 52% on reproduction). Nital
etch
a
tr
<
I
a.
o
o

0 0.050 0.100
DISTANCE FROM FUSION BOUNDARY-inches
Fig. 15 —Results of hardness traverse across the regions in Fig. 14

Fig. 16 —Microstructure of 5 s stationary repair weld in normalized heat Fig. 17 — Microstructure of 5 s stationary repair weld in normalized heat
27. Nital etch, X75 (reduced 55% on reproduction) 27 with postweld tempering at IIOVF (593°C) for 1 h. Nital etch, X75
(reduced 55% on reproduction)

virtually impossible to choose welding welding. Thus, 10 X 4 X V2 in. (254 X Figure 13 compares the as-cast and
conditions which provide cooling rates 102 X 13 mm) specimens from each heat normalized microstructures at X250. Note
sufficiently slow to prevent the formation were normalized at 1750°F (954°C) for 2 the complete elimination of the continu-
of continuous networks of crack-sensi- h. ous networks of pearlite that results from
tive microstructures in the regions which
experience partial transformation to aus-
tenite. n Q
The hardness levels identified in these CC UJv UJ
F- r-
regions are far above those shown to be Q. O UJ
acceptable for avoiding hydrogen- 5 w1 z ->
induced cracking even with good low- M
% r - 0 CQ
hydrogen practice. Thus, it would be < UJ /
<
Z.
difficult, if not impossible, to make repair r/
Q.
welds in as-cast steels without the risk of CO z>
forming microcracks in the high-carbon
martensitic networks. In fact, hindsight
now suggests that the lack of quantitative
correlation between the acoustic-emis-
sion studies and identifiable hydrogen-
induced cracks is attributable to this phe-
nomenon.
Unfortunately, microcracks in these
hard brittle networks in the partially trans-
formed region can serve as nuclei for
slow crack growth even after postweld
heat treatments. Therefore, it appears ~I000°F
that the only sure way to avoid this Fig. 19—Schematic of various regions produced in steel casting repair welds: FZ—fusion zone;
problem would be to eliminate the net- UMZ — unmixed zone; PMZ—partially me/ted zone; HAR — homogeneous austenite region; HAZ—
works of interdendritic pearlite nodules heat-affected zone; CG—coarse grained; E.G.—fine grained; PTR — partially transformed
by a homogenizing treatment prior to region

102-s | APRIL 1985

the normalizing treatment. Conclusions partially-transformed region.
Figure 14 shows a panorama of a weld 4. The problem in the partially-trans-
made in a normalized specimen from The following conclusions pertain to formed region arises from the interden-
heat 0. Comparison with Fig. 4 indicates the hydrogen-induced cracking studies of dritic segregation present in the as-cast
that the coarse continuous networks of repair welds in steel castings. material.
high carbon martensite have been elimi- 1. The metallurgical factor was con- 5. Normalizing eliminates the problem
nated by the homogenization treatment. trolled by using the method described in the partially-transformed region by
Furthermore, the fine grain size produced previously to select combinations of homogenization.
by normalizing lowers the hardenability welding conditions which produced both 6. All steel castings should be normal-
and helps to provide more acceptable crack-sensitive and crack-insensitive mi- ized before repair welding to minimize
microstructures. crostructures in typical case steel speci- problems in the partially-transformed
mens.
Figure 15 is a plot of the microhardness region.
transverse run on the weld shown in Fig. 2. The chemical factor was controlled 7. As the alloy content is increased,
14. Comparison with Fig. 12 clearly indi- by modifying the moisture content of the postweld-heat treatment may become
cates the advantage gained by normaliz- E7018 electrode coatings. desirable.
ing. Although these plots are for welds 3. The mechanical factor was synthe-
made with identical conditions, the maxi- sized by applying a controlled aug- Acknowledgments
mum hardness in the normalized material mented strain to SMA welds deposited
is only about 350 KHN as compared to on simple rectangular specimens. The authors wish to thank Mr. Dan
about 1200 KHN in the weld made in the 4. Acoustic-emission techniques were Stone and the Association of American
as-cast material from the same heat. employed to monitor the initiation and Railroads (AAR) for its financial support of
growth of hydrogen-induced cracking. this project.
Figures 16 and 17 show microstruc-
tures of welds made in normalized heat 5. The results confirm the fact that
proper low-hydrogen practice is manda- References
27 with and without a postweld temper
at 1100°F (593°C) for 1 h. The weld in tory for all repair welding operations if 1. Rogers, H. C. 1956. The influence of
Fig. 16 was not tempered and showed hydrogen-induced cracking in the coarse- hydrogen on the yield point in iron. Ada Met.
some networks of martensite with hard- grained heat-affected-zone is to be 4(2):114-117.
ness levels as high as 45 Re. However, avoided. 2. Savage, W. F., Nippes, E. F., and Homma,
6. The susceptibility to hydrogen- hi. 1976. Hydrogen-induced cracking in HY-80
after the postweld temper (Fig. 17), the
induced cracking is directly related to steel weldments. l/Ve/c//r?g,/oiv/7?,3/55(11):368-s
structures appear more homogeneous, to 376-s.
and the region of maximum hardness both the maximum hardness in the heat-
3. Frohmberg, R. P., Barnett, W. )., and
consists of tempered martensite with a affected zone and the level of diffusible Troiano, A. R. 1955. Delayed failure and
hardness of 40 Re. Figure 18 shows this hydrogen. hydrogen embrittlement in steel. Trans. ASM
same structure at X500. 7. Hydrogen-induced cracking can be 47:892.
The various zones, which are pro- prevented by using good low-hydrogen 4. Adams, C. M., |r. 1958. Cooling rates and
duced in the weldment during repair practice if the maximum hardness in the peak temperatures in fusion welding. Welding
heat-affected zone is kept below Re 35. lournal 37(5):210-s to 215-s.
welding of steel castings, are summarized
8. Since the maximum hardness in the 5. Savage, W. F., Nippes, E. F., and Sze-
in schematic form in Fig. 19. It is apparent keres, E. S. 1976. Hydrogen-induced cold
that the region most susceptible to heat-affected_ zone is directly related to
cracking in a low-alloy steel. Welding lournal
hydrogen-induced cracking is the partially the carbon content, reduction in the
55(9):276-s to 283-s.
transformed region; here networks of carbon content is the simplest way to 6. Ball, D. )., Cestal, W. J., Jr., and Nippes,
hard and brittle constituents are pro- improve the weldability of steel cast- E. F. 1981. Determination of diffusible hydro-
duced in the inhomogeneous austentite. ings. gen in weldments by the RPI silicone-oil extrac-
It is apparent from the above results The following conclusions pertain to tion method. Welding Journal 60(3):50-s to
the results of microstructural studies of 56-s.
that a normalizing treatment should pre-
repair welds in steel castings: 7. Savage, W. F„ and Lundin, C. D. 1965.
cede any repair weld of steel castings if
1. For cast steel repair welds one can The varestraint test. Welding lournal
the danger of hydrogen-induced cracking 44(10):433-s to 442-s.
is to be minimized. In general, with identify seven distinct regions by a nital
etch. 8. Aidun, D. K„ and Savage, W. F. 1984.
increases in both the alloying content and Optimizing repair welding techniques in cast
the section thickness, the segregation 2. The partially-transformed region of steels-Part I. Welding lournal63(11):345-s to
becomes more severe and ultimately a the heat-affected zone of repair welds in 353-s.
postweld heat tempering will become as-cast steel was found to have the 9. Savage, W. F., Nippes, E. F., and Sze-
necessary. microstructure most susceptible to keres, E. S. 1976. A study of weld interface
hydrogen-induced cracking. phenomena in a low alloy steel. Welding
3. Continuous networks with hard- lournal 55(9):260-s to 268-s.
ness levels ranging from 600 to 1200 KHN
(50 to 70 Re) were identified in the

WELDING RESEARCH SUPPLEMENT 1103-s