Preventing Stress Corrosion

Cracking in the Carbon Dioxide

Absorber of Ammonia Plants

Through field tests and experience, the threshold below which SSC
does not occur is estimated at 100 MPa. A new local stress relief
technique can lower residual welding stresses in that level.

Hiroshi Ishimaru and Tetsuya Takegawa

ABSTRACT field tests with precracked C-T

specimen to decide threshold stress
After nine year operation of CC>2 level below which S.C.C. does not
removal tower using Vetrocoke process, occur,
stress corrosion crackings (S.C.C.)
were found in 19?6 only on its locally The suitable local stress relief was
stress relieved circumferential butt applied to 101-E tower successfully,
welds. Though there was no difference and no S.C.C. took place for the
in hardness between locally and wholly following one year.
stress relieved welds, the residual
stresses of locally relieved welds
were far higher than that of wholly
relieved ones. By heating the limited 1. CC>2 removal tower
band width of shell, it bulged and
non-sustained bending moment was The 750 ton/day ammonia plant of
generated. This moment deformed the Sumitomo Chemical Company
heated portion of shell plastically, designed by M. W. Kellogg came on
and consequently, residual stresses stream in Niihama in May, 196?
were redistributed when cooled. To Potassium carbonate solution
lower that moment, the uniform temper- activated by arsenous oxide,
ature band width should be kept wide namely, Giammarco-Vetrocoke
and temperature gradient be minimized. solution, is used in CC>2 removal
process of this plant. It is
After carrying out intensive examina- reported that though carbon steel
tions with a large size model, we has a high corrosion resistance
determined the necessary uniform band against this solution to save
width and gradient of temperature to special alloy, it is susceptible
lower the residual stress below the to stress corrosion crackingl)2).
prescribed level. e also carried out
The C02 removal tower is made of
ASTM A212 Gr B and 45 m (148 ft)
in height, and consists of four
Sumitomo Chemical Company, Limited parts, each of which was wholly
Niihama, Ehime, Japan stress relieved in a closed
furnace before assembling in a
0149-3701-80-2959 $01.00 1980
American Institute of Chemical Engineers
170
vendor's shop. It has three We measured the inner surface
locally stress relieved circum- residual stresses of the tower by
ferential butt weld joints. e X-Ray method.
call them C-7, C-13 and C-l8.
(Weld lines were designated from As shown in Figure 1, the residual
C-l to C-23 from the top.) stresses of locally stress
relieved welds were over 20O MPa
The nominal composition of (29.0 ksi) which were extremely
Vetrocoke solution is 300 kg/m3 higher than those of wholly heat
K2C03, 110 kg/m3 AS23 and the treated ones.
ratio of C02 to K2 is in a range
from 1.7 to 0.9- Composition of After the cracks were ground off
process gas at the inlet is 52% and welded for repair, local
H2, 21% C02 and 17% 2 The stress relief was done according
temperature is about 110C in the to the ASME code in 1976.
middle of the tower. Repaired circumferential weld
lines were heated in uniform from
Details on stress corrosion outer surface with wrapping heater
cracking elements. Although the residual
stresses were still high and
Stress corrosion crackings were stress corrosion crackings took
found only on the three locally place again after a year opera-
stress relieved welds, namely, C-71 tion.
C-13 and C-l8 after nine year
operation of the tower. Before In 1977, those cracks were
that time, these welds had not repaired again and local stress
been inspected. relief was done with a longer
holding period and a higher hold-
The cracks, mainly branched ing temperature. Since more than
intergranular, are produced in the 20O MPa (29 ksi) of residual
deposit metal and propagated a stresses still remained, occur-
little into the base metal rence of stress corrosion cracking
longitudinally. The ratio of could not be prevented.
crack length on the metal surface
to depth was small. We measured residual stresses of
another C2 removal tower which
The Vickers hardness number of was made recently. The result
cracked welds is from l80 to 190 indicates that the levels of
and have no difference in hardness residual stress in wholly stress
from shop stress relieved ones. relieved welds are about 35 MPa
(5.1 ksi) while those of locally
To inspect these cracks from out- stress relieved ones are in a
side on streamly by angle beam range from 100 to 200 MPa (l4.5 to
technique of ultra-sonic, the 29 ksi). We concluded that under
depth of crack should be deeper the conditions that meet minimum
than 1/4 of wall thickness and requirements of ASME code, the
weld reinforcement of both surface local stress relief could not
be ground off. lower residual stress to the level
below which stress corrosion
Measurement of residual stresses cracking does not occur. The
of C02 removaltower by X-Ray result quantitatively certifies
method the notice by the British Standard
1515 Part 1 that local stress
The occurrence of stress corrosion relief is not effective to prevent
cracking is limited to the locally stress corrosion cracking.
heat treated welds. This is
because, we suppose, there is a On the other hand, it seems that
difference in the degree of relief the metallurgical benefit of
of residual stress between wholly softening the deposit metal and
and locally heat treated welds. the heat affected zone by temper-

171
ing hardened structure is produced. stress of material. There-
fore, this method is hardly
Stress level for safe operation applied to the tower.

The relation between value of (3) Flame plastic spraying

surface residual stresses of the
welds of our tower 101-E measured Carbon steel strips whose
by X-Ray method and occurrence of surface were sprayed with
stress corrosion cracking showed epoxi or fluorine resin were
that no stress corrosion cracking tested in the field. The
takes place below the stress of former was attacked by the
l60 MPa (23.2 ksi). Vetrocoke solution, and the
latter was not damaged by
The threshold stress intensity the same. The technology for
factor K value obtained through flame spraying of fluorine
field test with self loaded resin in the field have not
compact tension specimens is shown been established.
in Figure 2. The harder the
material is, the lower becomes the (4) Type 304 stainless steel
KISCC value. Contact of the strip lining
material with zinc, a less noble
metal, gives effect to prevent the The results of field test
precrack of the specimen from with U-Bended specimens show
propagating. that the type 304 stainless
steel does not have suscep-
The threshold nominal stress level tibility to stress corrosion
can be calculated from these cracking. But the large
results. The calculation accuracy difference in coefficient of
largely depends upon the estimated thermal expansion between
size of initial flaw. The thresh- carbon steel of shell and
old stress level necessary for stainless steel of lined
safe operation is decided to be metal redistributes the
100 MPa (14.5 ksi) considering residual stress at the fillet
those test results and field edge welds around all four
experience (a stress corrosion sides of lined metal during
cracking has not been taken place the post weld heat treatment.
on wholly stress relieved welds.)
(5) Effect of hardness to the
Comparison on methods for prevent- occurrence of stress corro-
ing stress corrosion cracking sion cracking

(1) Shotpeening Bent-beam specimens with

various hardnesses controlled
This is a method to obtain by heat treatment were tested
high compressive residual in the field for five years
stress most easily on the and checked by microscope
steel surface. In this occurrence of stress corro-
method, however, it is sion cracking on their cross
difficult to peen long weld sections. The threshold
line uniformly and the ten- hardness of normalized
sile residual stress comes ferritic-pearlitic structure
peak at the end of peened below which stress corrosion
parts . cracking does not occur is
somewhat higher than that of
(2) Vibrational residual stress martensitic or tempered
relief martensitic structure. These
results indicate that the
For sufficient stress relief, carbon steel, the hardness of
the vibrational stress is which is below 20O Vickers,
required to exceed the yield immune to the stress corro-

172
 sion cracking. Although the buldges locally and, as a result,
hardness of the tower is not non-sustained force is generated
higher than l80 Vickers , at the boundary between uniform
stress corrosion cracking and not-uniform parts of longitu-
takes place. dinal temperature distribution.
As the simplest example, a
We also studied the repair cylindrical shell with symmetrical
method in which the tower is linear temperature distribution to
welded from inside while weld line is assumed (Refer
cooling it from outside. It Figure 3) Then the maximum moment
seems difficult, however, to value Mo is generated at point 0
apply the method to the plant. and can be calculated as follows J"
Consequently, we began to
seek for the following new (t0 -
Mo = -( (1)
techniques for field heat
treatment to relieve residual
stresses below the threshold The moment MX at the weld line
level . which is apart from point O by X
is given by the following equa-
Examination on holding time and tion :
temperature
MX = -D ( d Wi
o'
Holding time and temperature d X^
necessary to relieve welding
stress below the threshold level
were studied with two types of (2)
small specimens : the one is
circular grooved weld and the Where a: radius of shell
other is weld overlayed by
electrodes to repaired C02 removal h: thickness of shell
tower .

Post weld heat treatment was made D:

in a muffle electric furnace, and 12(1 -
surface residual stresses were
measured by X-Ray method before
and after the heating. So it is =
clearly indicated that almost all
welding stresses are relieved after
heat treatment for an hour at 500C \j = Poison's ratio
(l,040F). The effect of tempera-
ture to stress relief is not E = Young's modulus
improved over 580C (1,O76F). In
order to prevent lowering of As yield strength of material
tensile strength, the material drops at high temperature, moment
temperature must not be high for a MX deforms weld line plastically
long time. and reproduces and/or redistrib-
utes residual stresses. On the
The reason why residual stresses other hand, Figure 11 given by
are not relieved by local heat S. T. Timoshenko indicates that
treatment the moment MX comes to almost 0 if
distance x is over 3/- This
We consider the residual stress is means if uniform temperature width
not relieved by local heat treat- is 6/, external force is not
ment under the conditions that meet applied to the weld line. It is
minimum requirements of ASME code impracticable to heat uniformly
because of the following: the band width of 6/ of actual
tower with a large radius, and
When circumferential band width of reproducing of stress at point 0
cylindrical shell is heated, it can not be avoided. So, we have

173
 carried out several experiments and heated from inside with space
with large size cylindrical shell heater units installed therein.
model to find applicable condi- Its outside is only insulated with
tions for successful local stress asbestos clothes. The temperature
relief to the tower. of inner surface is somewhat
higher than that of outer one, but
8. Local stress relief heat treatment as shown in Figure 7, the residual
test with a large size cylindrical stresses are relieved sufficiently.
shell model This method is concluded to be
most practical in the field.
The shell used in the test is
1.5 m (4.9 ft) in diameter, 3 m Judging from these results, it
(9.8 ft) in length, 32 mm (1-1/4 becomes clear that if the uniform
in) in wall thickness of ASTM temperature band width is kept
SA516 Gr 7O and has three circum- over 3/ it is possible to lower
ferential butt welds which are not the residual stresses in the weld
stress relieved. The residual below 100 MPa (14.5 ksi). And if
stresses were measured by X-Ray the distance .between the weld line
method. Each of three circum- and the point where the tempera-
ferential butt weld lines were ture comes a half of the maximum
heated from outside by wrapping value is larger than 3/i the
heater elements. The longitudinal residual stresses reproduced at
temperature distributions are the discontinuous point 0 of the
given in Figure 4. temperature distribution do not
come to the harmful level.
The residual stress values after
local stress relief by the conven- 9- Results of field heat treatment
tional temperature distribution f0rCO2 removaltower
CASE A which meets the ASME codes
are still high as experienced in We found out the method for field
the field tower. Use of low stress relief that can bring
strength welding rod as overlaying satisfactory result.
metal is effective to decrease
welded residual stresses, but it In March, 1978, we applied it to
gives no effect to the residual the locally repair welded 101-E
stress after stress relief. C02 removal tower. The tower was
heated by Bulk Head method. The
Figure 5 shows the longitudinal longitudinal distribution of
residual stress distributions temperature and the residual
before and after stress relief by stresses after heating at C-l8
CASE B with more gradual tempera- weld line are shown in Figure 8.
ture distribution than CASE A. The residual stresses in weld line
And results of CASE C with flat were relieved sufficiently below
temperature distribution which has the prescribed level to prevent
concave region at its center are stress corrosion cracking. There
shown in Figure 6. The width of was no difference in stress
uniform temperature region is distribution circumferentially.
2.3/- The residual stresses are The reproduced stress at the
relieved satisfactory. So minimum discontinuous point of longitu-
requirement of uniform temperature dinal temperature distribution
band width is determined. Since also did not exceed that level.
the tower has many holes, nozzles
and other devices, it is difficult After a year operation, inner
to add sufficient heating elements surface of the tower was inspected
thereto to keep wide and uniform by magnaflux and microscopic
temperature band width. We tested replica method in March, 1979-
the Bulk Head Method. In this No crack was found at the welds
method, the shell is partitioned of tower. It is still in a sound
off with bulk heads which face state at present.
each other across the weld line,

174
Conclusion Acknowledgement

(1) Stress corrosion crackings We wish to express our thanks to

were found only in the locally Mr. Kamohara, Hitachi Research
stress relieved welds of C(>2 Laboratory, Mr. Fukujouji, Mr. Matsuo
removal tower operated for and Mr. Morita, Kasado Works of
nearly nine years. The Hitachi, LTD. for their informative
Vetrocoke process was used, suggestion and work for experiments
but inhibitor for stress with large sized model and also to
corrosion cracking was not Messrs, the Kasado Works and Hitachi
used. Research Laboratory of Hitachi, Ltd.,
and Sumitomo Chemical Company for
(2) As a whole, the residual giving us permission to publish this
stresses at the welds article.
relieved in a closed furnace
are low enough, but they
exceed 200 MPa (29.0 ksi) at
the locally stress relieved Literature cited
ones. The residual stresses
were measured by X-Ray method. 1. Atkins, K,T,G., Fyfe, D. and
Rankin, J,D
(3) Residual stresses at the Corrosion in C2 removal plant
locally heat treated welds towers.
were not relieved. This is Presented at AICHE symposium on
caused by the moment safety in ammonia plants and
generated at the discontinu- related facilities, Vancouver,
ous point of longitudinal Sept. 1973-
temperature distribution. By
keeping the uniform tempera- 2. Naito, K., Hashimoto, T. and
ture band width across the Kihara, T
weld line over 3/i residual Stress corrosion cracking of mild
stress can be lowered below steel in gas treating plants.
the threshold level. J. Japan Petrol. Inst. 14 , 762
(1971)
(4) The threshold stress below
which stress corrosion crack- 3- Timoshenko, S,P. and Krieger, S.
ing does not occur is esti- Woinowsky
mated to be 100 MPa (l4.5 ksi) Theory of plates and shells,
through field tests and chap. 15, second edition, McGRAW-
experience. Materials with HILL
hardness higher than 200
Vickers have susceptivity to 4. Burdekin, F,M
stress corrosion cracking. Local stress relief of circum-
ferential butt welds in cylinders.
(5) Local stress relief which Brit, welding journal. Sept. 1963,
meets above mentioned 483-490.
requirements was conducted on
C2 removal tower 101-E, and
residual stresses were
confirmed to be below the
threshold level.
T* P
After a year operation, its
inner surface was inspected
and no stress corrosion
cracking was found at the
locally stress relieved
welds. ISHIMARU, H. TAKEGAWA, T.

175
 Weld (In* Residual tress (MPa) TOO,
40 80 120 160 200 "D MXM CASE A

C- 7 Deoo o MM CASC a
MM CASt C
HAZ .
Q
"V ^ 101 t 1*77
o 101 E 1*71 C-7
A
X ^X."'l E I97SC-C3
> C-13 Depo A

A
o
k. -^ ~ - - - - - -^p
- - - "

se
*
* e
^
A
j^ 200
:::::":::-:::::::::::::A l/ft Z/O 3/tt 4/fl s/a
il Another nflw
HAZ

-o
Length from

Figure 4. Longitudinal distribution of temperature at the

weld line

Dpo
101 E lower o
iC- o holding period of local stress relief heat
= C-22
JE ": Another new
HAZ :::::...<> treatment.
S IOIE fower D*po :~7A
O Stress relieved at Tenders shop.
-0 Stress relieved in 1976 in the field.
& Stress relieved In 1977 in the field. C-I weld line
Figure 1. Surface residual stresses of CO2 removal towers -6000
IO-E after local and whole stress relief heat
treatment.
IOO -500

(ksi) o>
-400
^ 60 60 100
80 Q.
E
N 50
-300
50-
40- O
O

28"
-100
I 20- Va 0 Vfl 2/fc SP 4//T
Length from C-l weld line
10- Figure 5. Distribution of residual stresses along the
longitudinal direction of large-sized model,
after local stress relief by Case B.
500 600 TOO
Tensile strength (MPa)
Figure 2. Relations between KjgCC *n G-V solution and C-3 W.U

tensile strength of carbon steels.

Temperature
0

-co-
2/rt l/f 0 l/f 2/f3 3/fl 4/13 *>/(*
Length from C-2 weld line

Figure 6. Distribution of residual stresses after local stress

Figure 3. Simple model of causing of bending moment by relief by Case C.
local heating of shell.

176
 -Weld line Bottom -weld line Top

M O \/ft 2/a V/9 5/a

Length from weld line

Figure 7. Distribution of residual stresses after local stress 3/rt 2/a 1/4 O 2/0 3//3
Length from weld line
relief by Bulk Head method.
Figure 8. Longitudinal distribution of outer surface
temperature and inner surface residual
stresses of Tower 101-E.

177