The Fatigue Resistance of Plasma and

Oxygen Cut Steel

The fatigue resistance of A572 steel flame-cut surfaces is
greater than for A514 steel at lives greater than 2 X 10
5
cycles but less at lives under IO
5
cycles
BY N-J HO, F. V. LAWRENCE, JR., AND C. J. ALTSTETTER
ABSTRACT. Oxygen-cut specimens of
controlled roughness, of gouges, and of
gouges repaired by welding and by grind-
ing were prepared for low-alloy, high-
strength steel (ASTM A572) and
quenched and tempered steel (ASTM
A514). A572 plasma-cut specimens were
also prepared. Load-controlled fatigue
tests were performed using a compres-
sion-to-tension stress cycle (R = 1).
Subsurface microstructures were charac-
terized by metallography and by micro-
hardness measurements.
The fatigue resistance of the flame-cut
surface was greater for A572 than for
A514 at lives greater than 2 X 10
5
cycles.
For A572 steel, the fatigue lives were
found to be somewhat different from
specimens having different cutting
methods. For A514 steel, the machined
surfaces were superior at lives greater
than 10
4
cycles. Heat treatment of the
A514 flame-cut surfaces did not much
improve their fatigue resistance.
Small gouges on the cut surface were
found to have negligible influence on the
fatigue resistance compared wi th the
flame-cut surface. However, deeper
gouges had a large negative influence on
the fatigue resistance: gouges repaired by
welding did not have any better fatigue
resistance. Grinding out the gouges did
not improve the fatigue resistance of the
A572 welds.
Introduction
Influence of Oxygen Cutting on Fatigue
Oxygen cutting and plasma-arc cut-
ting have been widely used in industry
because they are inexpensive, fast
methods of cutting complex shapes from
plate stock. In oxygen cutting a preheat
flame precedes the oxygen jet, but the
major heat input is due to oxidation of
the metal. Proper cutting conditions
suitable preheat flame, cutting flame and
cutting speed depend on the gas pres-
sure, nozzle type and type of gas. In
plasma-arc cutting, the heat is supplied by
a jet of superheated nitrogen or argon-
plasma which impinges on the metal. The
fatigue properties of oxygen-cut surfaces
may be altered as a result of changes in
chemical composition, microstructure,
residual stress and geometrical features
such as roughness, gouges, drag lines and
melted edges.
Several investigations on the effect of
flame cutting (Ref. 1-5) have shown that
the fatigue resistance and other proper-
ties of cut surfaces are quite variable due
to the use of different materials and
cutting conditions. Only the resulting sur-
face roughness has been systematically
controlled. The fatigue resistance of the
oxygen-cut surface has been found to be
strongly influenced by roughness, and
differences of as much as an order of
magnitude in life have been found
between rough and smooth surfaces at
the lowest stress levels (Ref. 3).
In practice, flame-cut surfaces of very
little or very great roughness are not
usual, since they are neither practical nor
economical. A reasonably good commer-
cial product might average 0.005 in. (0.13
mm) peak to valley. However, large
gouges produced by bl ow-out, lateral
torch instability, etc., are often encoun-
tered; these defects may drastically lower
the fatigue resistance. When gouges are
encountered, the alternatives are: rejec-
tion of the gouged part, grinding the
gouge to smooth and remove the sharp
notch, filling the gouge wi th wel d metal,
or leaving the gouge as is. The effective-
ness of each of these alternatives is not
clear.
Scope of Investigation
The principal objective of the present
paper was to evaluate the effect of
oxygen cutting on the fatigue resistance
of industrially important steels. High-
strength, low alloy (ASTM A 572) and
quenched-and-tempered (ASTM A 514)
steels were investigated at intermediate
At the writing of this paper, N-J Ho was a
Research Assistant, F W. LAWRENCE, JR. was
Professor of Civil Engineering and Metallurgy,
and C. I. ALTSTETTER was Professor of Metal-
lurgy, University of Illinois at Urbana-Cham-
paign
lives (less than 10
6
cycles). Surface rough-
ness was controlled to 0.0054 0.0010
in. (0.14 0.025 mm), peak to valley, in
order to compare the fatigue resistance
of different steels and different cutting
methods at the same surface roughness.
Surface gouges of various depths were
introduced into specimens and were ei-
ther left as formed, ground off, or filled
with wel d metal to examine the influence
of such conditions on the fatigue resis-
tance.
The fatigue behavior of oxygen-cut
specimens was compared wi th that of
plasma-cut specimens and specimens
having machined test surfaces. Several
tests were performed on specimens heat
treated after oxygen cutting. The micro-
structures and hardness profiles were
determined for each steel and cutting
method.
Experimental Procedures
Two structural steels one compara-
ble to ASTM A572 grade 42 (0.22% max.
carbon, 42 ksi min. yield strength), and a
Table 1Chemical Composition, Wt-%
<a)
A572 A514
C
Mn
Si
Cr
Mo
B
S
K
p
( a)
Composi ti on
0.22
1.35
0.30
-
-
-
0.05
0.04
suppl i ed by manufacturer.
0.179
1.165
0.263
0.032
0.151
0.002
-
-
Table 2Mechanical Properties
A572
A514
S
y
, ksi
Su, ksi
ef, %
RB
Rc
52-58
75-77
26
78

110
120
15.5
-
27
WELDI NG RESEARCH SUPPLEMENT 1231-s
Table 3Specimen Preparation
Series Steels Surface condition
1
2
3
4
5
6
7
A572 and A514
A572 and A514
A514
A572
A572 and A514
A572 and A514
A572
Machined
Flame-cut
Quenched-and-tempered after cutting
Plasma-arc cut
Gouged
Gouged, repaired by welding
Gouged, repaired by grinding
nor mal t o t he f l ame cut surf ace wer e
det er mi ned usi ng a Vi ckers i ndent er and
a 100 g l oad. Fatigue test speci mens wer e
mount ed in an MTS machi ne wi t h self-
al i gni ng gri ps and wer e f at i gued using a
compr essi on- t o- t ensi on stress cycl e
(R = - 1 ) under l oad cont r ol at 3- 5 Hz.
Af t er f at i gue t est i ng, t he f r act ur e surfaces
wer e exami ned wi t h a scanni ng el ect r on
mi cr oscopy (SEM).
quenched- and- t emper ed steel whi ch
conf or med t o ASTM A514 grade C
(0. 20% max. car bon, 0. 005% max. bor on,
100 ksi mi n. yi el d st rengt h) we r e
obt ai ned as 1 i n. (25.4 mm) t hi ck pl ates.
The chemi cal composi t i on and t he
mechani cal pr oper t i es of t hese steels are
gi ven in Tabl es 1 and 2. The series of test
speci mens t hat wer e pr epar ed is shown
in Tabl e 3.
Al l speci mens wer e cut wi t h t he l oad-
i ng di r ect i on per pendi cul ar t o t he rol l i ng
di r ect i on. Speci mens of series 1 t o 4 wer e
cut t o t he di mensi ons shown in Fig. 1.
Because of t he gouges, t he speci mens of
series 5 t o 7 wer e cut t o t he di mensi ons
shown in Fig. 2. Speci mens of series 2 and
3 wer e f l ame cut f r om bl anks by movi ng
t hem beneat h a cut t i ng t or ch whi ch was
osci l l ated at const ant ampl i t ude (0. 0054
i n. , i.e., 0. 14 mm, peak t o val l ey) and at a
const ant f r equency by a cam mechani sm
(Ref. 3). The cut t i ng par amet er s (Tabl e 4)
wer e caref ul l y cont r ol l ed t o mi ni mi ze
resol i di f i cat i on deposi t s on t he mel t ed
edge (Ref. 3). A t or ch t i p t o met al separa-
t i on of 0.25 i n. t o 0.5 i n. ( 6. 4- 12. 7 mm)
was used.
Af t er cut t i ng most speci mens, t he hot -
rol l ed surfaces wer e machi ned t o a 0. 01
i n. (0.25 mm) dept h keepi ng t he t ool
marks paral l el t o t he l oadi ng di r ect i on.
Series 3 was subsequent l y heat t r eat ed t o
obt ai n t he same mi cr ost r uct ur e and har d-
ness as t he quenched- and- t emper ed
steel t hr oughout its thi ckness. Speci mens
Flame Cut Surface
in series 4 wer e pl asma-arc cut by Cat er -
pillar Tr act or Co. using a ni t r ogen pl asma
gas and a cut t i ng speed of 30 i pm (12.7
mm/ s) . Half of t hese speci mens had t hei r
hot - r ol l ed surfaces r emov ed by machi n-
i ng.
For Series 5, gouges of vari ous dept hs
wer e made by suddenl y movi ng t he
t or ch l ateral l y or l etti ng a t hi n sheet of
steel pass t hr ough t he cut t i ng j et. The
cut t i ng condi t i ons wer e t he same as se-
ries 2 and 3 except t hat t he t or ch was not
cont i nuousl y osci l l at ed. In series 6, t he
gouges wer e cl eaned and r epai r ed by
wel di ng usi ng E6013 cover ed el ect rodes
f or A572 and E11018 cover ed el ect rodes
f or A514 and t hen mi l l ed smoot h. Speci -
mens of series 7 had t he gouges
r emoved by gri ndi ng t hem t o a gener ous
radius (3 i n. , i.e., 76. 2 mm) , t o a dept h just
suffi ci ent t o r emove t he gouge.
All cut speci mens wer e wi r e br ushed
by hand bef or e surf ace cont our prof i l i ng
and t est i ng. The cont our of t he f l ame-cut
surface and gouge of each speci men was
measur ed usi ng an LVDT pr of i l omet er
(Ref. 3) wi t h a t ungst en car bi de stylus
i ncl i ned 30 deg f r om t he nor mal surf ace.
Profi l e traces (Fig. 3) wer e r ecor ded at
t he upper edge, mi dt hi ckness and l ower
edge t o det er mi ne if t he speci mens wer e
wi t hi n t ol er ance (0. 0054 0. 0010 i n.
peak t o val l ey average roughness).
Sel ect ed speci mens of each series
wer e sect i oned and mount ed f or opt i cal
met al l ogr aphy. Mi cr ohar dness prof i l es
a 2.0 R
c
a
-
3.75"
^
I.75"
I.O'
I.C
r-
)" I.75"
N
0.25"R \ ,
3.75"
m-
3.0
J
X
I.O
-Machined Surface ^Fl ame Cut Surface
Fig. 1 One inch wide specimen geometry
Results
Metal l ography and Microhardness
Flame cut t i ng pr oduces an al t er ed
mi cr ost r uct ur e and vari at i ons in mi cr o-
hardness. Al l f l ame- cut speci mens had a
l arge heat - af f ect ed zone at t he t or ch
(upper) si de of t he speci mens due t o t he
larger heat i nput f r om t he cut t i ng nozzl e.
The heat - af f ect ed zones of t he upper
edges are t wo t o t hr ee t i mes deeper t han
t hose of t he l ower ( bot t om) edges of t he
speci mens.
A572 Flame-Cut Specimens. The
mi cr ost r uct ur es at t he mi dt hi ckness of
t he f l ame- cut surf ace r esembl e t hose
r epor t ed by Gol dber g (Ref. 4) and t he
Net herl ands Gr oup (Ref. 5). At t he cut
surf ace, a 25- 50Mm thi ck car bon-
enr i ched zone consi st ed of spots of l ede-
buri t e on t op of a t hi n l ayer of pearl i t e,
f ol l owed by pl at e mart ensi t e, mi xed wi t h
ret ai ned aust eni t e.
The pl at e mart ensi t e and r et ai ned aus-
t eni t e wer e al ways f or med at t he surf ace
of t he r oot of serrati ons behi nd t he
cut t i ng f l ame. The hardness of t he pl at e
mart ensi t e and ret ai ned austeni te mi xt ur e
was over 800 VHN (Fig. 4). The f i ne
pearl i t e under t he l edebur i t e was ar ound
500- 450 VHN (not appar ent in Fig. 4).
The mi cr ost r uct ur e at t he peak of t he
serrat i on due t o cut t i ng consi st ed mai nl y
of vari ous amount s of f i ne pearl i t e gr ow-
i ng at aust eni t e boundar i es wi t h l edebur -
ite spot s at t he t i p of t he serrat i on.
Bel ow t he f usi on line t o a dept h of 0.5
mm (0.02 in.) was a zone cont ai ni ng l ath
mart ensi t e and/ or bai ni t e. To a dept h of
0.4 mm (0.016 in.) bel ow this zone was
an i nhomogeneous mi xt ur e of pearl i t e
and fi ne f er r i t e. At an addi t i onal dept h of
0.4 mm (0.016 in.) a subcri ti cal heat -
af f ect ed zone consi st ed of parti al l y sphe-
r oi di zed pearl i t e and f er r i t e. Finally, ori gi -
nal pearl i t e and f erri t e subst r uct ur e was
f ound.
A572 Plasma-Arc Cut Speciments. The
mi cr ost r uct ur e of t he pl asma-arc cut sur-
f ace at t he l ower edge, is di f f er ent f r om
t hat of oxygen- cut speci mens. Ther e was
no evi dence of sel ecti ve oxi dat i on or
car bur i zat i on pr oduced by a pl asma gas
j et. Ther e was no har d car bur i zed l ayer
t o be seen.
At t he cut surf ace, t her e was a mar t en-
site matri x t o a dept h of 0.3 mm (0.012
mm) bel ow t he surf ace. However , wi t hi n
this zone, t he mart ensi t e at t he cut sur-
f ace was less har d t han t hat bel ow t he
232- s | NOVEMBER 1981
cut surface (Fig. 5). This effect may be
due to loss of carbon from the molten
material. Under the martensite matrix
zone was a fine ferrite plus high carbon
austenite. Beyond this was ferrite and
very small spheroidized pearlite adjacent
to the original pearlite-ferrite structure.
A514 Flame-Cut Specimens. At the
midthickness of the cut surface, the
microstructure and microhardness (Fig. 6)
for the A514 oxygen-cut specimens was
similar to the A572 oxygen-cut speci-
mens. Ledeburite, plate martensite,
retained austenite and pearlite at the cut
surface, and lath martensite matrix zone
below the fusion line were found. They
extended to a 0.5 mm (0.02 mm)
depth.
For the A572 steel, a hard layer and
high carbon region were found. Further
below was coarse-grain tempered mar-
tensite, then there was a layer of many
small pearlite nodules spreading at the
grain boundaries. This coincided wi th the
abrupt drop of microhardness (see Fig. 6)
and probably represents the intercritical
heat-affected zone.
A514 Quenched-and-Tempered After
Cutting Specimens. There was a straight
boundary between light and dark-etching
regions beneath the cut surface. This
straight boundary was believed identical
to the fusion line seen on the as-cut
surfaces. Within the light region at the
top there were many spheroidized fine
carbides suggesting that carbon was
picked up during cutting and was sphe-
roidized upon subsequent heat treat-
ment.
Fatigue Test Results
The results of the fatigue tests are
plotted in S-N diagrams, Figs. 7 to 11.
Some specimens failed at undesired loca-
Flame Cul Surface
t
Gouge - Fl ame Cul Surface
Fig. 2 Two inch wide specimen geometry
tions and hence were not included in the
S-N diagrams. The lines in Figs. 7 to 11
were fitted to the test data using the
least-squares method.
Discussion
Fatigue Crack Initiation Sites
Fatigue crack initiation of the flame-cut
A572 specimens occurred at the edge of
the flame-cut surface. For the flame-cut
A514 specimens, initiation occurred ei-
ther close to the upper edge or at the
midthickness of the cut surface. The
A514 usually had multiple points of initia-
tion. In A514 specimens, quenched-and-
tempered after cutting, the fatigue cracks
initiated at points on the flame-cut sur-
face, mainly at the most severe
notches.
All flame-cut surface failures were at
the roots of serrations. It seems that
geometry is more important than micro-
structure in determining the fatigue crack
initiation site. For the plasma-arc cut A572
specimens, the fatigue crack initiation
point was mainly at the upper edge, as
Goldberg reported (Ref. 4). In those
specimens with unground hot-rolled sur-
faces, the crack always initiated on the
hot-rolled surfaces probably as a result of
surface decarburization, rolling defects,
or unfavorable residual stresses.
Influence of the Flame-Cut Surface
Roughness
The geometry or surface roughness of
the flame-cut surface is one of the most
important factors which control fatigue
resistance. A model for the effect of the
flame-cut surface was developed by
assuming that surface contour can be
approximated as a series of notches (Ref.
3). The fatigue notch factor Kf is defined
as the ratio of fatigue strength of
unnotched to notched specimens at long
lives and is usually less than the theoreti-
cal stress concentration factor K
t
(which is
defined as the ratio of the maximum
longitudinal stress at the tip of the notch
to the average, remote stress) because of
plasticity effects. Peterson (Ref. 6) found
that these effects can be empirically
accounted for by the relationship:
K
f
= 1 + K, - 1
(1)
1 -Fa_
r
where r = radius of the notch; a = ma-
terial property.
For steel, the value (a) can be related to
the ultimate strength (S
u
):
a = 0.001(300/S
U
)
18
(2)
Fig. 3 Profilometer traces of (A) flame cut
surface, (B) plasma cut surface, and (C) gouged
surface
C - - & - - Q_o C
0.2 0.4 06 OS 1.0 1.2 1.4
Di st ance Below Surf ace (mm)
Fig. 4Microhardness profile at the midthickness of A572 flame-cut steel
J L
WELDI NG RESEARCH SUPPLEMENT 1233-s
Table 4Oxygen Cutting Parameters for 1
i n. Steel Plate Using Propylene Gas
(a)
Table 5Comparison of Calculated and Measured Fatigue Stress Concentration Factor
A572 A514
Gas pressures; psig:
Propylene
Preheat oxygen
Cutting oxygen
Average cutting speed, ipm
""Cut t i ng t i p - " Ox we l d 14Z38."
wher e (s
u
) is gi ven in ksi and (a) is in
i nches.
As can be seen f r om t he surf ace pr o-
files of Fig. 3 and t he i deal i zati ons of Fig.
12, t he roughness of t he surf ace may be
appr oxi mat ed as a series of semi ci rcul ar
not ches ( even t hough t he not ches mor e
nearl y r esembl e semi el l i pti cal not ches).
Unf or t unat el y, t he K
t
of a series of semi -
elliptical not ches is not avai l abl e. The K
t
f or a single el l i pti cal not ch is:
K
t
= 1 + 2 V D / . ( W/ 2 )
(3)
wher e D = dept h of t he not ch;
W = wi dt h of t he not ch.
The K
t
of a semi el l i pti cal not ch at t he
cut surface is smaller t han t hat of a
semi ci rcul ar not ch, because VV is great er
t han D, yet it was assumed t hat K
t
f or a
series of semi el l i pti cal not ches wi l l be
r oughl y t he same as t hat f ound f or semi -
ci rcul ar not ches. Usi ng t he model f or a
series of semi ci rcul ar not ches, K
t
val ues
wer e det er mi ned (Ref. 7) usi ng average
val ues of D and D / W f or each gr oup of
speci mens Tabl e 5.
Ul t i mat e st r engt h (S
u
) val ues wer e esti -
mat ed f or each steel f r om hardness dat a.
The f at i gue not ch f act or , K
f
, was t hen
cal cul at ed usi ng eqs. (1) and (2). The
val ues cal cul at ed are shown i n Tabl e 5
and compar ed wi t h exper i ment al l y det er -
mi ned val ues. The machi ned speci men
results wer e t aken as t he smoot h speci -
men f at i gue st r engt h at t he val ue of stress
whi ch caused fai l ure in I O
6
cycl es.
The cl ose agr eement in Tabl e 5
bet ween measur ed and cal cul at ed val ues
Material VHN S
u
, ksi D, in.<
c
> D/W<
c
>
Kf
calc.
Kf
meas.
3.5
6
48
14
4.5
7
50
14
A572
(flame-cut)
A514
(flame-cut)
A514 (Q&T)
390
560
280
180<
a
>
280<
a
>
120
(b
>
0.00251
0.00112
0.0052
0.0056
0.0053
0.0060
0.081
0.081
0.081
1.556
1.665
1.431
1.0
1.60
1.43
<a>
Estimated f r om hardness measurements averaged f r om surface t o 0.25 mm (0.01 mm) dept h
t b)
From tensile test.
t c )
Averaged values.
of Kf f or A514 suggests t hat t he di f f er-
ences bet ween machi ned and f l ame-cut
results can be at t r i but ed t o geomet r y and
surf ace roughness ef f ect s al one. Thi s is
not t he case wi t h t he A572 results. Possi-
bl y resi dual stress has had mor e i nf l uence
on A572 t han on A514.
Residual Stress
Surface resi dual stresses al ter t he mean
stress whi ch, in t ur n, great l y i nfl uences
t he f at i gue crack i ni ti ati on l i fe. The stress
ampl i t ude, mean stress l evel and cycl es t o
fai l ure are rel at ed t hr ough t he rel at i on
(Ref 8):
N
f
=
[^P]
(4)
wher e Nf = cycl es t o f ai l ure; cr
a
= stress
ampl i t ude; o\ = f at i gue st rengt h coef f i -
ci ent ; <r
0
= mean stress; b = f at i gue
st rengt h exponent (negat i ve val ue).
Compr essi ve resi dual stresses r educe
t he mean stress l evel so t hat t he f at i gue
life is l onger, and t he f at i gue st rengt h is
hi gher. Tensi l e resi dual stresses shor t en
t he f at i gue l i fe. The Net herl ands Gr oup
(Ref. 5) s howed this ef f ect f or f l ame- cut
surfaces. However , t hese surf ace resi dual
stresses onl y pl ay i mpor t ant rol es in t he
f at i gue crack i ni t i at i on, and are pr esumed
t o have little ef f ect on t he pr opagat i on
stage due t o large crack t i p pl asti ci ty
whi ch el i mi nates any resi dual stress. The
resi dual stresses on t he f l ame-cut surface
depend on cut t i ng par amet er s in a ver y
compl i cat ed way ; even wi t h t he same
o o o p
o o
J I I L
Di st ance Below Sur f ace (mm)
Fig. 5Microhardness at the lower edge of A572 plasma-cut steel
cut t i ng condi t i ons, t he resi dual stresses
are never exact l y t he same f r om speci -
men t o speci men.
The resul ts of resi dual stress measur e-
ment s by t he Caterpi l l ar Tr act or Co. usi ng
x-ray di f f r act i on met hods i ndi cat ed t hat
t he l ongi t udi nal resi dual stresses in one
speci men of A572 steel wer e ent i rel y
compr essi ve on t he f l ame- cut surf ace,
whi l e in a speci men of A514 steel small
tensile resi dual stresses wer e obser ved
on one surf ace Fig. 13. From eq. (4), t he
f at i gue life of A514 shoul d be less
i nf l uenced by t he resi dual stress t han t he
A572 whi ch has a smaller C'I.. These
compr essi ve resi dual stresses of t he same
l evel shoul d have a st ronger i nf l uence on
A572 t han on A514. This is t hought t o be
t he reason wh y t he measur ed K
t
val ue is
smaller t han t he cal cul at ed Kf val ue f or
A572 steel .
The Effects of Gouges
Whet her gouges have a st r ong i n-
f l uence on t he f at i gue st r engt h depends
on t he gouge dept h and t he stress con-
cent r at i on f act or caused by t he gouge.
The wi dt h of t he gouge cannot be
smaller t han 0.125 i n. (3.2 mm) , t he di a-
met er of t he cut t i ng j et . Thus, it is di f f i cul t
t o make ver y sharp gouges by cut t i ng. It
is not pract i cal or possi bl e t o make ver y
deep gouges by suddenl y movi ng t he
cut t i ng nozzl e l ateral l y wi t hout l osi ng cut -
ti ng act i on.
For bot h steels i nvest i gat ed, small
gouges ( appr oxi mat el y less t han 0.02 i n. ,
i.e. 0. 51 mm, dept h average) and some
large gouges (0.06 i n. , i.e., 1.5 mm aver-
age) di d not r educe t he f at i gue st r engt h;
fai l ure i ni ti ated on t he f l ame- cut or hot -
rol l ed surfaces. For t he A572 steel , onl y
t hree speci mens fai l ed at gouges whi ch
had dept hs over 0.02 i n. (0. 51 mm) aver-
age; but even in t hese cases, t he f at i gue
st rengt h was r educed onl y slightly c om-
par ed wi t h t he f l ame-cut speci mens of
t he same di mensi on wi t hout gouges. For
A514 steel , onl y t hose speci mens wi t h
gouge dept hs over 0.06 i n. (1.5 mm)
fai l ed at t he gouge. Ot her gouge speci -
mens fai l ed at t he f l ame- cut surfaces and
had f at i gue lives ver y cl ose t o t hose of
t he f l ame- cut surf ace wi t hout a gouge.
Gouge- wel ded speci mens of A572
steel whi ch f ai l ed at t he wel d had some-
what poor er f at i gue behavi or t han unr e-
pai red gouged speci mens (Fig. 8), yet
234- s | NOVEMBER 1981
02 0.4 06 0.6 1.0 | 2 1.4
Distance Below Surface (mm)
Fig. 6 - Microhardness at the midthickness of A514 flame-cut steel
A572 R=- l
o Machined
A Flame Cut
Plasma Cut
I _L I J I I I I I I
Cycles To Failure
Fig. 7S-N diagram for 1 in. A 572 steel specimens
A572 , Flame Cut R=- l
Without Gouge
V Gouge, Failed At Gouge
o Gouge Repaired By Welding
Gouge Repaired By Grinding
I I I I I I I I I
-I I I I M i l l
Cycles To Failure
Fig. 8S-N diagram for 2 in. wide A572 steel specimens; gouge depths as noted
A5I 4 R = - l
o Machined
A Flame Cut
0 Heal Treated 1 0 * 1 ]
J I I I I I I I I I 1 I I I I I I I I I I I I I I l l
Cycles To Failure
Fig. 9S-N diagram for 1 in. wide A514 steel specimens
t hose of t he A514 whi ch fai l ed at t he
wel d r oot s had a poor er f at i gue resis-
t ance t han gouged speci mens (Fig. 10).
The reason f or this is t hought t o be t hat ,
whi l e repai r wel di ng t he deeper gouges
in t he A514 steel , compl ex resi dual
stresses may have been f or med and wel d
def ect s such as l ack-of -f usi on and slag at
t he wel d r oot wer e i nt r oduced.
Gouges coul d be r epai r ed by gri ndi ng,
t oo, if t he gouge is not ver y deep. The
results showed that gri ndi ng does not
seem t o have benef i ci al ef f ect s f or t he
A572 steel .
Comparison of Flame-Cut, Plasma-Cut and
Machined Surfaces
As seen in Fig. 7 f or A572 speci mens
wi t h 1 i n. (25.4 mm) wi dt h, t her e is onl y a
small di f f er ence bet ween t he f l ame- cut ,
pl asma-cut and machi ned surfaces. It
seems t hat t he pl asma-cut surf ace has a
smoot her surf ace and a little hi gher f at i -
gue resistance t han t he ot her s. As seen in
Fig. 10 f or 2 i n. (50.8 mm) wi de gouge-
gr ound speci mens, fai l ure al ways oc-
cur r ed at t he f l ame-cut and hot - r ol l ed
surfaces. None of t hese f ai l ed on t he
gr ound surf ace ( machi ned surface). It
seems that t he f at i gue resi stance of t he
f l ame-cut surface depends on t he ef f ect s
of residual stress, geomet r y and, t o a
mi nor ext ent , mi cr ost r uct ur al al t erat i on.
Several speci mens wi t h t he ori gi nal
hot - r ol l ed surf ace f ai l ed at t he surf ace.
The f at i gue results showed t hat t he hot -
rol l ed surf ace had t he same f at i gue resis-
t ance as a machi ned surf ace (Fig. 11).
From this resul t, it woul d seem t hat it is
not necessary t o r emove t he hot - r ol l ed
surface.
The super i or i t y of t he A514 machi ned
surfaces is seen in Fig. 9. Most of t he
f l ame-cut and quenched- and- t emper ed
speci mens fai l ed at t he f l ame-cut surf ace
ot her t han t he machi ned surf ace on t he
lateral sides. However , all t hr ee t ypes of
surfaces gi ve t he same f at i gue resistance
at short er lives (10
4
cycl es). Compar i ng
Fig. 11 S-N diagram for 1 in. wide A572 steel
specimens with unground hot rolled surfaces
-
_

-
-
' I
D a
A 5 I 4 Fl ame Cut R - - I
A Wi t hout Gouge
o Gouge, Not Foi l ed At Gouge
Gouge Fai l ed At Gouge
a Gouge Repai red By Wel di ng
11 i l
0.037
0. 025
I
'
rOOIS"
A ^ ^ - ^ ^ j . 0 017'
0.06=T~"----^
0* ^i f cr j OC
/

09
" f oie
1
Hj oe"
1 1 1 1 1 if
'
.

-0.02"
' 0.03 " " " - " " - ^ ^
^^^
i i i i i i i 1
Cycles To Failure
Fig. 10 S-N diagram for 2 in. wide A514 steel specimens
Fig. 12 Actual surface roughness (A) and
idealized semicircular notches (B)
WELDING RESEARCH SUPPLEMENT 1235-s
10 10
Cycles To Failure
Fig. 14 Comparison of A572 and A514 flame-cut surfaces fatigue resistance
j L J
Flome Cut Surfoce Flame Cul Surfai

Torch Side
Flame Cut Surf
\
k Flame Cul Surface
Fig. 13 Distribution of longitudinal residual
stress at the surface (ksi): A -A572; B-A514
quenched-and-tempered and the as-cut
surfaces, the heat treatment did not
much improve the fatigue resistance. At
10
6
cycles, the quenched-and-tempered
specimens showed a 3 ksi (20.7 MPa)
fatigue strength improvement over the
flame-cut specimens.
Effects of Different Dimensions and
Materials
In Fig. 14, the fatigue resistance of
flame-cut specimens wi th different
widths is shown. Wider specimens had a
lower fatigue resistance for both steels.
On the average, the fatigue strength was
decreased 5 ksi for 2 in. (50.8 mm) wide
specimens compared wi th the 1 in. (25.4
mm) wide. All specimens were in plane
strain condition so that this effect may
result from differences in residual stresses
along the cut surface caused by different
preheat and cooling rates for the differ-
ent widths.
In Fig. 14, it is seen that A572 had
higher fatigue resistance of the flame-cut
surface than A514 beyond 2 X 10
6
cycles. Previous work (Ref. 3) also
showed this tendency beyond 10
6
cycles
for the smooth surface and beyond
2 X 10
5
cycles for the rough surface at
stress ratio, R = 0. The A514 exhibited a
greater notch sensitivity than A572.
Conclusions
1. For the A572 steel, the differences
in the fatigue resistance resulting from
different cutting methods are very small.
For the A514 steel at lives greater than
1.5 X 10
4
cycles, the machined surface
has greater fatigue resistance than the
flame-cut and quenched-and-tempered
(after cutting) surfaces. Heat treatment of
the flame-cut surface does not much
improve fatigue resistance.
2. The difference in fatigue resistance
between flame-cut and small gouged sur-
faces is negligibly small for both steels. Big
surface gouges have a negative influence
on the fatigue resistance. Neither grinding
nor repairing gouges by welding
increases the fatigue resistance com-
pared with the gouged surface for both
steel.
3. At lives greater than 2 X 10
5
cycles,
A572 flame-cut surfaces have greater
fatigue resistance than A514, but at lives
less than 10
5
cycles the fatigue resistance
of A514 flame-cut surfaces is greater than
that of A572.
Acknowledgments
This work was sponsored by the
Caterpillar Tractor Company, Peoria, Illi-
nois. The suggestions and cooperation of
Mr. Willis Fildes and Mr. Cal Loyd of that
company were important contributions
to this work. The use of the testing
facilities of the Civil Engineering Depart-
ment of the University of Illinois are also
appreciated.
References
1. Koenigsberger, F., and Farcia-Margin, Z.
1965. Fatigue strength of flame-cut specimens
of bright mild steel. British Welding journal,
January: 37-41.
2. Koenigsberger, F., and Green, H. W.
1955. Fatigue strength of flame-cut specimens
in block mild steel. British Welding journal, July:
313-321.
3. Plecki, R.; Yeske, R.; Altstetter, C; and
Lawrence, F. V., |r. 1977. Fatigue resistance of
oxygen-cut steel. Welding lournal 56 (8): 225-s
to 230-s.
4. Goldberg, F., 1973. Influence of thermal
cutting and its quality on the fatigue strength of
steel. Welding lournal 52(9): 392-s to 404-s.
5. Netherlands Institute of Welding. 1973.
The properties of flame-cut edges, Final report
of Working Group 1913, May.
6. Peterson, R. E. 1959. Notch sensitivity.
Chapter 13 in Metal Fatigue, eds. Sines and
Waisman. McGraw-Hill Book Col., Inc.
7. Peterson, R. E. 1974. Stress concentration
factors. New York: Wiley and Sons, Inc.
8. Graham, ]. A. 1968. Sec. 3.2 in SAE
fatigue design handbook, ed. Morrow,
loDean. Society of Automotive Engineers.
236-s | NOVEMBER 1981