Engineering Failure Analysis 13 (2006) 10761091

www.elsevier.com/locate/engfailanal

Failure analysis of pipes used in a hydrodesulfuration

system of a petrochemical plant
C. Barbosa *, S.M.C. de Souza, R.O. Centeno, I.C. Abud, O.B. Ferraz
Instituto Nacional de Tecnologia (INT), Avenida Venezuela, 82, Rio de Janeiro, 20081-312, Brazil

Received 8 July 2005; accepted 14 July 2005

Available online 9 September 2005

Abstract

Some types of petroleum have high sulfur content, depending on the region of extraction. In order to make possible
the industrial use of such oil, it is necessary to subject it to a hydrodesulfuration process. In this process, water is used to
extract sulfur from the oil, and thus, generating acid water, which is carried away through a pipe system. In this work,
failures in a pipe system were analyzed through optical microscopy and Vickers microhardness measurements. Just the
most signicant part of the several weld joints of this enormous hydrodesulfuration system is presented here. The fail-
ures were associated to the aggressive environment, weld defects, which acted as stress raisers, and an inadequate micro-
structure which increased cracks propagation.
2005 Elsevier Ltd. All rights reserved.

Keywords: Sulde stress corrosion; Hydrogen embrittlement; Welding; Microstructure; Failure analysis

1. Introduction

Hydrodesulfuration systems are employed in reneries whenever the sulfur content of the processed oil is
high, due to geological reasons. In this kind of installation, water is used to catch sulfur from the oil yield-
ing an acid water, which is carried away through a pipe system. Usually this pipe system is composed by
welded high strength low alloy steel tubes subjected to the harmful action of a very aggressive environment
with high amounts of H2S. The early failure of components by fracture/cracking in H2S containing envi-
ronments is considered to be the result of stress corrosion induced by sulde (SSC). The main factors which
contribute to the occurrence of this failure mechanism can be divided in two main groups: environment

*
Corresponding author. Tel.: +55 21 2123 1203; fax: +55 21 2123 1284.
E-mail address: cassiob@int.gov.br (C. Barbosa).

1350-6307/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2005.07.014
 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091 1077

Table 1
Chemical composition (mass %)
Section C Si Mn P S Ni Mo Al V Nb Ti Pb Fe
1 0.20 0.19 0.83 0.015 0.007 0.01 0.01 0.02 0.001 0.002 0.001 0.002 Bal.
2 0.18 0.20 0.52 0.013 0.009 0.01 0.02 0.02 0.001 0.001 Bal.
3 0.13 0.23 1.30 0.017 0.005 0.01 0.04 0.02 0.045 0.034 0.002 Bal.

Fig. 1. General aspect and welded regions of section 1.

(H2O, concentration of H2S, pH, temperature and stresses) and materials (yield strength, microstructure,
heat treatments and alloying elements). The presence of H2S has a signicant eect, causing hydrogen
embrittlement in steel. Besides, H2S and sulde species slow down the formation of molecular gaseous
1078 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091

hydrogen (H2), favoring the formation of atomic hydrogen, which enters the steel crystal lattice. In some
systems ammonia is present in high levels. It reacts with H2S, producing ammonia bisulde, which gener-

ates NH +
4 and HS , inhibiting the recombination of H (mainly in high pH values), favoring hydrogen
embrittlement. A common feature of SSC and hydrogen embrittlement is the start of fracture on sites with
high local stresses. The dierence between each kind of mechanism lies on the presence of H2S in SSC.
Therefore, SSC is a special case of hydrogen stress corrosion, also known as hydrogen embrittlement [13].
Several factors can inuence SSC: temperature, partial pressure of H2S, presence of other gases, such as
O2 and CO2, species in water solution and time of permanence of water in the pipe [1,2]. The main factors
not related to environment are microstructural characteristics, which can favor cracking, and the presence
of weld defects, which act as stress raisers [3]. Stress relief heat treatment can minimize, but does not elim-
inate cracking completely, since it is associated to some microstructural features. Some hard and brittle
phases favor the occurrence of hydrogen embrittlement, generating cracks usually in the heat aected zone
(HAZ) and in the weld metal (WM), due to an inadequate thermal cycle [47]. Another important issue,
regarding the use of steels in H2S environment, is hardness, which can not be higher than 22 Rockwell
C (248 HV) [8].
In this work, optical microscopy and hardness measurements were used to analyze the failure of pipes
from a hydrodesulfuration system of a renery. This hydrodesulfuration system had several weld joints
which were investigated, but just the most signicant part of them is presented here. This analysis provided
evidences that an aggressive H2S environment associated to inadequate welding procedures which gener-
ated defects in the HAZ/WM interface, and to unfavorable microstructure, led to pipes failure.

2. Experimental procedure

The chemical composition of the steel used in the fabrication of the pipes is shown in Table 1. It complies
with a commercial ASTM A 106 grade B steel subjected to standard rolling procedures and heat treatment

Fig. 2. General aspect and welded regions of section 2.

C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091 1079

Fig. 3. General aspect and welded regions of section 3.

1080 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091

Fig. 4. (A) Macroscopic view of joint A of section 1 and their microstructures (B) and (C) a crack on the WM/HAZ interface on the
root of the weld joint. (D) Microstructure of two dierent weld regions: one with column shaped non recrystallized grains, typical of an
as solidied microstructure, and the other with more polygonal grains which were recrystallized by the following weld pass on the top
of the weld. (E) Detail of the recrystallized grains region.
 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091 1081

Fig. 5. (A) Macroscopic view of joint B of section 1 and their microstructures. (B) A crack on the HAZ/WM interface on the root. (C)
Bainiticmartensitic microstructure of the HAZ on the root. (D) Microstructure of the top weld with two regions: one with coarse
column shaped grains, resulting from solidication, and the other one with recrystallized grains, as result of following weld pass. (E)
Brittle bainiticmartensitic microstructure of the HAZ of the top weld region.
1082 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091

for this kind of application. The pipe system was assembled by tungsten inert gas (TIG) welding on the root
and by shielded arc electrode welding on the top of the weld joint.
Samples were cut from the weld bead and surrounding areas, subjected to standard metallographic
preparation and etched with a Nital 2% (nitric acid in ethanol) etchant. Vickers hardness measurements
were performed with a 98 N load on the weld metal (three measurements), heat aected zone (three
measurements on each side) and base metal (three measurements on each side).

Fig. 6. (A) Macroscopic view of joint C of section 1 and their microstructures. (B) A crack which originated at the heat aected zone
(HAZ)/weld metal (WM) interface on the root. (C) HAZ and weld zone also on the root. (D) Detail of brittle microstructure
(martensite and bainite) in the HAZ of the root.
 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091 1083

Fig. 7. (A) Macroscopic view of joint A of section 2 and their microstructures. (B) A crack propagating along the HAZ/WM interface
on the root. (C) Microstructure of the HAZ on the root, with presence of martensite and bainite. (D) Microstructure of the HAZ on
the top of the weld. (E) Detailed view of this microstructure with widespread brittle pro-eutectoid ferrite.
1084 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091

Fig. 8. (A) Macroscopic view of joint B of section 2 and their microstructures. (B) and (C) Microstructure of the HAZ on the root,
with some bainitic and martensitic regions. (D) and (E) A crack on the HAZ/WM interface on the root.
 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091 1085

Fig. 9. (A) Macroscopic view of joint A of section 3 and their microstructures, while cracks and porosity can be seen on a detailed
image. (B), (C) and (D) Cracks originated on the HAZ, but which propagated through the base metal.
1086 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091

Fig. 10. (A) Macroscopic view of joint B of section 3 and their microstructures. (B) and (D) A crack on the HAZ on a detailed view.
(C) Column shaped grains on the WM. (E) Martensiticbainitic microstructure of the HAZ.

3. Results

3.1. General overview

Fig. 1 shows the general aspect and welded regions of Section 1 of the pipe system, Fig. 2 presents the
same aspects relative to Section 2, Fig. 3 shows Section 3.
 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091 1087

3.2. Light microscopy

Figs. 410 show the microstructural characteristics of the welded joints documented in Figs. 13. Fol-
lowing several features, found in many joints, are listed:

 porosity in the WM on the top of the joint (Fig. 9);

 cracks originated at the heat aected zone (HAZ)/weld metal (WM) interface on the root of the weld
(Figs. 48);
 martensiticbainitic brittle microstructure in the HAZ and HAZ/WM interface on the root and top of
the weld (Figs. 4,5,7,8,10);
 column shaped non recrystallized grains typical of solidied microstructure in the WM (Figs. 4 and 5);
 regions with more polygonal grains, which were recrystallized by the following weld pass on the top of
the weld also in the WM (Figs. 4 and 5);
 brittle pro-eutectoid ferrite in both the HAZ at the top of the weld and in the boundaries of the coarse
columnar grains on the WM (Fig. 7).

Fig. 11. Diagram with regions where hardness was measured.

Table 2
Hardness values (HV 10)
Region/ Section 1 Section 1 Section 1 Section 2 Section 2 Section 3 Section 3
points joint A joint B joint C joint A joint B joint A joint B
BM left
1 145.54 151.42 195.08 168.47 176.88 165.39 169.87
2 140.79 158.79 191.25 154.95 170.88 169.54 170.64
3 156.08 164.02 205.36 153.66 159.78 155.02 169.79
HAZ left
4 172.39 180.14 237.61 172.11 183.38 205.80 202.07
5 194.87 175.41 237.39 166.50 186.52 235.53 233.18
6 183.28 202.33 263.05 188.15 194.93 250.36 242.66
WM
7 187.94 213.69 235.33 189.47 194.98 245.38 218.08
8 202.21 200.42 194.55 187.75 204.31 271.54 209.79
9 186.47 198.03 215.75 209.91 175.13 219.26 200.66
HAZ right
10 182.81 191.46 244.85 202.06 173.98 233.12 202.04
11 165.29 158.74 200.45 171.68 162.91 194.22 189.64
12 161.33 155.02 184.69 193.06 153.79 193.30 187.69
BM right
13 140.99 140.94 142.08 174.06 148.85 158.73 164.07
14 134.57 143.73 158.44 193.30 147.83 158.71 158.98
15 151.24 136.82 153.74 166.73 149.08 159.91 168.45
1088 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091

Fig. 12. Hardness proles, joints A, B and C of section 1.

3.3. Hardness

Fig. 11 shows the regions and individual points where the hardness measurements were done. Table 2
presents the hardness values obtained in the weld joints of the analyzed sections.
Figs. 1214 show the hardness proles for each section joint.

4. Discussion

The macroscopic and microstructural analysis the weld joints led to the conclusion that some factors
associated with the aggressive environment, caused the failure of the pipe. One of them is the presence
of several weld defects, such as:
 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091 1089

Fig. 13. Hardness proles, joints A and B of section 2.

 Misalignment: geometrical mismatch between the two welded parts, which are leveled o, generating
harmful defects and high local stresses.
 Inadequate reinforcement on the weld bead (root and top) with an excessive pile up of weld metal, which
outstands from the base metal.

The discontinuities generated by these defects act as stress raisers, and become preferential sites for the
nucleation of cracks, even when there is no aggressive environment, but becoming critical in the presence of
embrittling agents such as H2S [13]. It was observed that cracks were generated on the HAZ/WM inter-
face, where the inadequate reinforcement formed real notches, which acted as stress raisers. From these cor-
ner aws, intergranular cracks, typical of hydrogen embrittlement, propagated through the thickness of the
pipe wall, from the inner side (root of the weld joint) towards the external surface (top of the weld joint) of
the pipe wall.
The cracks propagation was favored by a brittle microstructure on the HAZ of the weld joints, which
comprised martensite, bainite and pro-eutectoid ferrite [47]. Usually these unfavorable microstructural
features are consequence of inadequate welding conditions, such as welding speed, cooling rate and other
parameters. Another disadvantageous microstructural factor is the presence of coarse column shaped
grains in the WM of the top of the weld, which is also related to the cooling conditions after welding [4].
Therefore, there are necessary and sucient factors that can cause hydrogen embrittlement or stress cor-
rosion cracking, which are: susceptible microstructure, internal stresses and a source of hydrogen, i.e., H2S
that ows inside the pipes. This fact was conrmed by the origin of all cracks always on the root, which is
placed on the inner side of the pipes.
1090 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091

Fig. 14. (A) Hardness prole, joint A of piece 3. (B) Hardness proles, joints B of piece 3.

The hardness results show that the HAZ is much harder than the base metal (BM), and its hardness is
closer to the one of the WM and, therefore, they are in accordance with the presence of hard and brittle
phases (martensite and bainite) already shown in the metallographic analysis.

5. Conclusion

After analyzing the results obtained by using optical microscopy and hardness tests techniques and con-
sidering all the available information about this subject, one can say that the cause of the pipes failure was
hydrogen embrittlement/sulde induced stress corrosion cracking due to the presence of an aggressive envi-
ronment, which contains H2S acting upon a welded structure with important defects. These were misalign-
ment/mismatch and inadequate reinforcement with lateral concavities on the root, which generated stress
raisers, from which cracks nucleated and propagated through a brittle microstructure (martensite, bainite
and proeutectoid ferrite) present on the HAZ/WM interface of the weld joints.

References

[1] Kane RD, Cayard MS. Roles of H2S in the behavior for engineering alloys, a review of literature and experience; Corrosion NACE
98, San Diego, March 2227, 1998, paper 274, p. 128.
[2] Jelink J, Dagata S, Bonis M, Louge MF, Crolet JL. Selection of C110 grade for mildly sour service, UK Corrosion and Eurocorr
94, vol. 3, Bournemouth, October 31November 3, 1994. p. 2432.
 C. Barbosa et al. / Engineering Failure Analysis 13 (2006) 10761091 1091

[3] Metals Handbook. Corrosion, 9th ed., vol. 13, ASM International, Metals Park, 1987. p. 14571.
[4] Tsay LW, Chen YC, Chan SLI. Sulde stress corrosion cracking and fatigue crack growth of welded TMCP API SL X65 pipeline
steel. Int J Fatigue 2001;23:10313.
[5] Balasubramanian V, Gulra B. Eect of welding processes on toe cracking behavior of pressure vessel grade steel. Eng Failure Anal
2004;11:57587.
[6] Carvalho MC, Deus EP. Fracture analysis of a ow control device used in the petrochemical industry. Eng Failure Anal
2003;10:4239.
[7] Tuma JV, Sedmark A. Analysis of the unstable fracture behavior of a high strength low alloy steel weldment. Eng Fracture Mech
2004;71:143551.
[8] NACE standards for materials requirements suldes stress cracking resistant metallic materials for oileld equipment, NACE MR
0175-94.