Proceedings World Geothermal Congress 2005

Antalya, Turkey, 24-29 April 2005

Hydrogen Induced Cracking of Low Strength Steels in Geothermal Fluids

Keith A. Lichti, David M Firth and Annette D Karstensen

MPT Solutions, PO Box 31-310, Lower Hutt, New Zealand
k.lichti@matperf.com d.firth@matperf.com a.karstensen@matperf.com

Keywords: Carbon steel, geothermal, stress corrosion immunity to Sulfide Stress Corrosion Cracking (SSCC) and
cracking, hydrogen embrittlement, fitness-for-purpose Hydrogen Embrittlement (HE). A caveat that has also been
present since before 1975 is the need for design stresses less
ABSTRACT than yield stress.
Typically, low strength steels and annealed stainless steels These guidelines were not simply “rules of thumb” but were
are specified for geothermal energy applications because of based on laboratory measurement and documented
the risk of Sulfide induced Stress Corrosion Cracking experience in the oil and gas industry, Milliams and Tuttle,
(SSCC) and Hydrogen Induced Cracking (HIC) in the 2003. The applicability of these “rules of thumb” was
presence of H2S containing fluids. Recent experience has demonstrated for geothermal applications for example by
demonstrated the risk of SSCC and HIC, sometimes known Marshall and Tombs, 1969. This demonstration was done
as Hydrogen Embrittlement (HE), of low strength steels for relatively low H2S concentrations that would normally
subject to high residual stress derived from fabrication not be of major concern to oil and gas industry materials
techniques. Unexpected cracking, discovered in two selection experts but in the absence of significant
geothermal pressure vessels, was attributed to HE cracking hydrocarbon it is believed that the standard should be strictly
in the welds. Fitness-for-purpose assessments completed on applied for any H2S containing geothermal environments.
the two vessels found one suitable for operation, with a
temporary repair, while a replacement vessel was being Two pressure vessels made using low carbon steel which
fabricated but the second required immediate removal from met the requirements of the NACE MR0175 standard and
service for repair. The cause of cracking was attributed to had hardness less than 22 HRC (248 HV) were found to be
use of submerged arc welding leading to high residual stress cracked after 2 and 3 years of service in a geothermal
in the welds of the 32 mm thick vessel walls. The vessels steam/brine environment and a geothermal steam
were made in compliance with the ASME VIII Div 1 design environment respectively. The cracks were associated with
code that allows vessels up to 32 mm wall thickness to be circumferential seam and strake welds in one vessel and
fabricated without Post Weld Heat Treatment. The vessels with nozzle welds in the other.
also met the requirements of NACE MR1075 for resistance
to sulfide stress cracking, having hardness less than HRC 22. In each case an assessment of the cracking propensity was
The experience gained suggests some general “rules of carried out: cracking distribution, depth of cracking, and
thumb” for avoidance of HE cracking in vessels used for metallograhic examination. Defect assessments defined
geothermal service: fitness-for-purpose to allow continued operation while
repairs and replacements were planned. A sample of
1. The number of weld passes must be as many as the wall cracked weld material was available from one of the vessels.
thickness in millimeters ie for a 32 mm wall at least 32 This paper describes the root cause of the cracking including
passes should be applied. the cracking distributions, engineering critical defect
2. All vessels should be stress relieved unless it can be assessments and material properties that led to the observed
proven that this is not necessary by the manufacturer cracking and procedures specified to repair the vessels.
for the welding procedure chosen.
3. Care should be taken to ensure any closing welds are
2. BRINE ACCUMULATOR AND STEAM PURIFIER
correctly stress relieved.
VESSEL CRACKING
In December 2002 severe cracking was detected in a brine
1. INTRODUCTION
accumulator in a geothermal power station after 2 years of
Historically, selection of materials for pressure vessels service, see Figure 1. In the following year, severe cracking
exposed to mixtures of geothermal steam and brine with was detected in a steam purifier of the same power station,
high levels of hydrogen sulfide have been required to see Figure 2. The station was commissioned in 2000. Both
comply with NACE International standard NACE MR0175 vessels were critical to the operation of the station.
(last published in 2003) in the same manner as for sour gas
environments in the petrochemical industry. This standard
has now been replaced by a joint NACE/ISO standard that is
published in 3 parts, NACE MR0175/ISO 15156-1:2001(E),
NACE MR0175/ISO 15156-2:2003(E) and NACE
MR0175/ISO 15156-3:2003(E).

The new joint international standard builds on guidelines

that were present in the previous NACE standard with
enhancements from European standards with similar
concerns, Milliams and Tuttle, 2003. A common premise
for the NACE standard which was initially released in 1975
has been that carbon steels having hardness less than
Rockwell C 22 (< 22 HRC) will in most cases have Figure 1: View of Brine Accumulator Vessel.
Lichti, Firth and Karstensen

The brine and steam are both expected to form iron sulfides
and iron oxides on the surface of carbon steel components
that block the corrosive environments from the metal
surface. The brine is expected to precipitate thin silica
scales over time.

The vessels experience a planned annual shutdown when the

station undergoes preventative maintenance and occasional
short duration planned and unplanned outages.

2.2 Vessel Design and Materials

The vessels were designed to ASME VIII Div 1.

Materials selection required compliance with NACE

MR0175 for both of the vessels. A plain carbon steel to
ASME SA 516 Grade 70 was used. Manway and relief
nozzles of the purifier were to SA 106 Grade B.

NACE standard MR0175 specifies, that for carbon steels in

the normalised and welded condition, that the steels should
Figure 2: View of Steam Purifier Vessel have a hardness lower than 22 HRC, (equivalent to 248 HV).
The standard does not specifically require a Post Weld Heat
2.1 Vessel Operations Treatment (PWHT) for stress relief unless the steel used has
been cold worked, reference is made to the requirements of
The brine accumulator takes separated water from two
ASME VIII Div 1 in this standard.
vertical separator vessels at a separator station located in the
steam field. Relatively short steam and brine pipelines take ASME VIII Div 1 pressure vessel code only requires vessels
the fluids from the separator station to the power station. with wall thicknesses greater than 1 ¼ inch (32mm) to be
The steam passes through the steam purifier located outside stress relieved. Both vessels were designed with 32 mm
the power house just before the steam turbine. wall thickness and were not stress relieved.
The temperature of the separated brine/steam in the brine ASME SA 516 Grade 70 has a specified minimum yield
accumulator is 212°C at saturation pressure, while the steam strength 262 MPa (typically 354-368 MPA) and a minimum
entering the purifier is at a temperature of 209°C and tensile strength of 483 MPa (typically 522-537 MPa). The
operating pressure of 19.5 barg. brine accumulator materials mill certificates indicated a
Charpy Impact at -40°C of 105-146 J. The vessel details are
The brine in the brine accumulator is a low chloride fluid,
given in Table 1. The purifier had an Type 304 stainless
around 2,700 mg/kg, with a low tendency for silica scaling
steel base.
at the temperatures encountered in the brine accumulator.
The brine is near neutral to slightly alkaline pH at
temperature 3. BRINE ACCUMULATOR DAMAGE ASSESSMENT
3.1 Cracking
The purifier is intended to remove dissolved and suspended
solids from the steam by a small amount of condensation, The brine accumulator is shown schematically in Figure 3.
encouraged by injection of cold condensate, the design The shell was made from five strakes welded to domed ends.
consists of internal screens and drains to achieve the Each strake was made from rolled steel (MR0175 allows hot
required purification. The steam/condensate had a partial rolled carbon steels) with a longitudinal weld that was
pressure of H2S of 0.25 kPa at a pH of 5.9 (neutral pH is located about ¾ the way up the vessel, these were identified
5.6). as S1 to S5 welds. The position of these was staggered from
one strake to the next on opposite sides of the vessel. Six
circumferential welds were present, C1 to C6.

Table 1 Pressure vessel design details.

Brine accumulator Steam purifier
Wall thickness (mm) 32 32
Diameter (mm) 2500 1981
Welding of longitudinal and Submerged arc to SFA,5.17, F7A2 –EM12K Not Available
circumferential seams
Weld preparation / procedure Double V with initial welding on the inside, air arc Not available
gouge from outside and then external welding
Post weld heat treatment None None
Test pressure 39 barg 41.2barg
Maximum design pressure 26 barg 26 barg
Normal working pressure 17 barg 23 barg
Operating temperature 193°C brine/203°C steam 222°C steam
Date of commissioning February 2000 February 2000
Previous inspection Minor cracks were found visually during a previous No cracking found
inspection. These disappeared during grinding previously
Fluid level in vessel 30% of volume None

2
 Lichti, Firth and Karstensen

Strake 5 Strake 4 Strake 3 Strake 2 Strake 1

14.6 m

Figure 3: Illustration of brine accumulator weld layout.

mixture of mill scale, welding slag from the submerged arc

welding process and corrosion product formed in service.
Cracks were evident in these deposits over a number of
welds.

Removal of the corrosion products and scales by grinding

revealed a greater distribution of cracks, Figures 5 and 6.
The majority of the cracks ran transverse to the welds and
appeared to be either centered about the centerline of the
weld or one of the heat affected zones.

A contracted NDT operator was engaged to find and

measure all of the cracks in the vessel, Table 2. Inspections
were carried out using black light magnetic particle
inspection and Ultrasonics. Cracking was seen in all 5 of the
Figure 4: The typical surface of the vessel with a fine horizontal longitudinal seam welds but only in a few of the
crack seen in the formed corrosion products circumferential welds. The distribution of the detected
and appearance after removal of corrosion cracks were reported as follows, Table 2:
products.
• The cracks appeared in discrete regions, they were not
evenly spaced along the entire length of the welds and
significant portions of the welds were crack free.
• The cracks ran either vertically or horizontally.
• All the cracks in the circumferential welds were close
to the junction with the adjacent longitudinal weld.
• The cracks were relatively straight and were not heavily
branched
• The cracks were open and easy to detect
• The deepest crack found was nearly through wall.

Table 2 shows:
• Number of cracks in vertical direction, perpendicular to
seam welds and parallel to circumferential welds.
• Longest and deepest vertical cracks.
• Number of cracks in horizontal direction, parallel to
seam welds and to perpendicular circumferential welds.
Figure 5: Vertical cracking in a longitudinal seam weld • Longest and deepest horizontal cracks.
as seen after removal of corrosion products.
During operation the water level was about ½ way up the No cracking was seen at the outside surface of the vessel in
vessel and a distinct water line could be seen with the lower the areas where insulation was removed for inspection of
portion having evidence of a different scale compared to the external welds.
top. The top portion of the vessel was covered with a layer
of oxide/corrosion product, Figure 4, presumed to be a
3
Lichti, Firth and Karstensen

The grains present in the capping welds were aligned

perpendicular to the surface with a band of ferrite along the
interdendritic grain boundaries. The weld had a typical
distribution of inclusions seen as round black particles, these
being variously rich in Mn/S, Fe/O and Al/O. The weld
structure was typical of being produced by submerged arc
welding with a high energy input. The cracking seen at the
surface was relatively straight, at time branched and the
cracks were open, Figures 4, 5, 6 and 7. The wide cracks
were filled with corrosion product indicating the cracking
had been present for some time.

The section of brine accumulator shell material provided

later had a horizontal seam weld with a typical crack
Figure 6: Horizontal crack transverse to circumferential extending into the parent material. A cross section of this
weld C5 seen after removal of the corrosion horizontal weld, Figure 8, also shows the results of a
products. Vickers Hardness survey and macro etching. The hardness
values were all less than 248 HV (< HRC 22). A total of 10
The welds had what appeared to be at least one large
passes were used for the weld, 6 on the outside and 4 on the
capping pass applied to both the inner and outer surfaces.
inside.
These had not been ground and typically the caps were about
3 mm higher than the adjacent plate. No significant welding Two parts of a large weld crack were cut from the provided
defects were detected. However, at least one weld repair sample. One was sectioned through a crack, the second was
was present. This repair had been applied at the time of broken open to reveal the fracture surface. The crack was
manufacture. open and filled with corrosion products, Figure 9. Three
major cracks initiated and one propagated to within 5 mm of
A survey of surface hardness was completed, Table 3. The
the outside wall. The crack directions, on the macro scale
average hardness measurements taken were all below 250
were all perpendicularly to the surface and were all
HV (< HRC 22) and there was no indication that the vessel
relatively straight. The majority of the cracking was
had not been made in accordance with the requirements of
transgranular. However, on a microscopic scale side
the NACE MR0175 or ASME VIII Div 1 standards.
branches going backwards and at 90 degrees to the main
A similar hardness survey was carried out in the adjacent cracks were in evidence.
steam separators, Table 4. Hardness values were up to 256
The weld was etched using Nital to reveal the microstructure
HV (< 23 HRC). However, no cracking was detected.
in the cracked areas, Figure 11. The cracks in the capping
pass on the inside of the vessel followed distinct ferrite areas
3.2 Metallurgy of Welds and Cracking in the capping weld that were present along the
Figure 7 illustrates the in-situ metallography of the brine interdendritic grain boundaries and these were subsequently
accumulator vessel walls in the area of cracking. Figure 8 corroded. This suggests that the cracks initiated
shows a cross section of a weld area that was provided later. intergranularly. The cracks that appear to be going in the
The parent plate shell material had a fine grained reverse direction were also corroded out areas of ferrite.
ferritic/pearlitic structure with a typical grain size of 0.01 The crack tip area of the shorter cracks were in a weld
mm. The welds had a coarse grained structure typical of a tempered zone containing finely dispersed iron carbide in
low carbon steel weld and the HAZ had a very coarse grain ferrite. The long crack ended in the capping weld of the
size up to 0.1 mm, with a coarse Widmanstatten structure. outside of the vessel.

Table 2: Summary of cracking seen in the brine accumulator vessel welds.

Strake/Circumferential Welds S1 S2 S3 S4 S5 C1 C2 C3 C4 C5 C6
Number of vertical cracks 26 7 5 24 33 0 2 0 0 0 0
Longest vertical crack (mm) 80 80 80 100 60 140
Deepest vertical crack 20 22 30 27 20 11
Number of longitudinal cracks 2 0 0 0 0 4 0 3 6 0
Percentage of weld crack free (%) 40 80 70 50 50 100 90 90 90 90 100
Longest longitudinal crack (mm) 100 20 85 80
Deepest longitudinal crack (mm) 7 10 25 25

Table 3: Results of internal surface hardness survey of the brine accumulator

Minimum Average Maximum
HV ( HRC) HV ( HRC) HV ( HRC)
Plate material 169 (<6) 182 (6) 193 (9)
Longitudinal weld 201 (11) 224 (17) 268 (25)
Longitudinal HAZ 197 (10) 207 (13) 234 (19)
Circumferential weld 200 (11) 220 (16) 238 (20)
Circumferential HAZ 207 (13) 217 (15) 227 (17)

4
 Lichti, Firth and Karstensen

Table 4: Results of internal surface hardness survey of one of the steam separators.
Average Maximum Minimum
HV (HRC) HV (HRC) HV (HRC)
Plate material 191 (9) 203 (12) 174 (<6)
Longitudinal weld 195 (10) 216 (15) 181 (6)
Circumferential weld 215 (15) 230 (18) 206 (13)
Circumferential HAZ 256 (23) 271 (26) 242 (20)

Figure 7: In situ metallography around accumulator cracking in the weld (on left) and the parent material (on right).

188 192
193 202 215
169 215 192 195 209 216 231 182
189 242

194 193 198 197

201 212 210 180

201 198 218 212

197 218
172 196 230 227 212 181

Figure 8: Cross section of typical horizontal seam weld in brine accumulator. Vickers Hardness (BS427: Part 1:1961)
results, HV10.

5
Lichti, Firth and Karstensen

Figure 9: Open crack in longitudinal seam weld filled with corrosion products.

Figure 10: SEM views of the crack tip, side branches and crack initiation sites.
The crack opening and crack tip areas were examined in • Crack mouth area: S/Fe in central area, Fe/S crystals
detail using a Scanning Electron Microscope (SEM) with within the S/Fe areas, a layer rich in Si/O, Fe/O next to
Energy Dispersive X-ray (EDX) analysis facility. The the metal surface.
results are summarized in Figure 10. There were no
remnants of the crack that had not been subsequently These analysis are consistent with products seen in similar
corroded. Corrosion products and scales seen in the cracks environments, namely magnetite (Fe3O4) in the crack tip and
were as follows: next to the metal surface at the crack mouth, a layer of silica
(in this instance separating the magnetite from the iron
• Crack tip areas: Fe/O with occasional islands of Fe/O/S sulfides and iron sulfides that were mainly pyrite (FeS2) with
some pyrrhotite (Fe(1-x)S).

Figure 11: Microstructure in areas of cracking; Top in weld metal, Bottom in parent metal.

6
 Lichti, Firth and Karstensen

3.3 Fitness for purpose assessment of brine accumulator The defects were, however, considered unacceptable for
long term operation as a result of the risk that the cracks
The vessel defects were of concern because of their size and could continue to grow to unacceptable levels. The vessel
the number of cracks prevented consideration of vessel replacement plans continued.
repair. However, it was desired to operate the vessel for a
period of time while a replacement vessel was being
fabricated. 3.4 Operation of the brine accumulator after the
assessment and replacement vessel fabrication changes
A fitness-for-purpose assessment was carried out using Following the defect assessment and the repair of the large
BS7910 Level 1 (and using mill certificate tensile and defect it was recommended that the vessel could be returned
impact property data) with a view to defining the risk of to service. This was approved by the statutory authority and
vessel bursting during operation. The transverse cracks in the power station was returned to service without significant
the longitudinal welds, at a maximum of 160 mm long and downtime. However, a condition of continued operation
27 mm deep were shown to be acceptable. In addition it was was that the vessel should be inspected on a monthly basis to
shown that the cracks would leak before burst. These cracks ensure cracks were not growing. The replacement vessel
had initiated in the weld and had only propagated a short was made of ASTM A516 Grade 70, using conventional
distance into the parent material suggesting that the cracking welding methods as outlined in ASME IX for V and Double
was associated with residual stress. The hardness of the V preparation multi pass-welds. Although the vessel wall
parent material was much less than 22 HRC indicated a was only 32 mm a full PWHT followed by hydro test was
strength less than 690 MPa. As a result, once the cracks completed. The specification for the new vessel was
enter the parent material crack propagation will decrease therefore modified from that of the original vessel in that:
especially as the residual stress will diminish.
1. A number of weld passes were required to ensure that a
However, the defects running perpendicular to the applied fine grained weld structure was obtained with some
hoop stresses in the vessel needed to be separately assessed degree of tempering of prior weld passes.
as these cracks remained in the HAZ and weld material 2. Post weld heat treatment was required.
where a residual stress equivalent to the yield stress of the 3. A maximum hardness of 250 HV (< 22 HRC) was
plate was assumed, 350 MPa. This stress was significantly specified.
greater than the applied pressure stresses of 100 MPa. The
deepest longitudinal cracking in the HAZ of the longitudinal The original repaired brine accumulator vessel was safely
weld was for 2 cracks in series, a 100 mm crack plus a gap operated for the time required to fabricate the replacement
of 70 mm plus a 70 mm crack with these being 7 to 8 mm vessel.
deep, giving an interacting defect 240 mm long and 8 mm
deep. This defect was shown to be unacceptable for service 4. STEAM PURIFIER DAMAGE ASSESSMENT
with the assumed high residual stress which is required for a
Level 1 assessment. 4.1 Cracking
The steam purifier is shown schematically in Figure 12. The
These cracks were repaired by preheating to 200oC for 1 base of the vessel was made of stainless steel while the
hour to remove hydrogen, then repair welded using a temper upper shell was made from 32 mm carbon steel as noted in
bead weld process and then subjected to crack testing after Table 1. The inlet and exit nozzles were reinforced while
48 hours. Minor surface cracking was found and shown to the manway and relief nozzles were not. All of the welds
be fit-for-purpose. After a hydro test at a pressure of 39 had more than one finishing pass on the inner surface.
barg the vessel was approved for operation by an
independent inspection authority.

D ef ec ts fo u n d i n th e re gi o n of the re lie f v al ve
I nl e t no z zl e, i nte r m itte nt
de f e c ts f o un d o n th e i ns i de
w e l d b e t w e e n s hell a n d O utl e t no z z le, inte rm i tte nt def e c ts
no z z le, f o u n d o n the i ns i de we l d b e t w e e n
s he ll a n d no z z le

F ull y c i rc u mf e re nti al
c rac k o n the i nsi de of
th e v e s se l w a s f o u n d
a ro un d the m a n w a y

Figure 12: Illustration of cracking distribution of steam purifier welds.

7
Lichti, Firth and Karstensen

Cracking was observed on the inside surfaces in the nozzle operator characterized the length and depth of the cracks as
to shell welds. The small diameter relief nozzle had shallow noted in Table 5.
radial cracks. The inlet and outlet nozzles had short cracks
along the circumference of the welds, as illustrated in Figure A survey of the inner surface hardness was completed on
13, while the manway had a longer circumferential crack three nozzles, Table 6. Hardness values were up to Vickers
that went right around the nozzle. A contracted NDT 248 (< HRC 22).

Table 5: Distribution of cracks found in the steam purifier.

Number of Longest Deepest Location/ Orientation Distribution
cracks (mm) (mm)
Inlet nozzle / Numerous 420 22 HAZ and in weld / Even around
shell weld (160 to 420 Parallel to weld circumference
mm)
Outlet nozzle / Numerous 280 15 Lack of fusion Around one quadrant
shell weld (35 to 280
mm)
Manway / shell 1 Continuous 22 HAZ in plate at toe of weld / Continuous around
weld lack of fusion circumference
Relief valve 6 61 19 Radial ie transverse to weld Star pattern around
nozzle

Table 6: Hardness survey average value results for three nozzle welds in the steam purifier.

Relief valve Manway Outlet nozzle

HB (Brinell) HV (HRC) HB HV (HRC) HB HV (HRC)
Parent (main shell) 149 144 (<1) 217 219 (15.4) 243 248 (22)
Weld 153 148 (<1) 165 161 (<1) 147 141 (<1)
Adjacent to weld 126 118 (<1) 142 136 (<1) 236 240 (20)

The cracking on the relief nozzle was similar to that seen in

the brine accumulator being perpendicular to the weld. The
cracking in the manway and the inlet nozzle was in the toe Nozzle
of the weld on the plate side, while the cracking in the outlet
nozzle was in the toe of the weld on the nozzle side, see
Figure 13. The longest flaw found on the inlet nozzle was
420 mm long and 22 mm deep.

Nozzle

Figure 13: Illustration of cracking at the top of the outlet

nozzle at the toe of the weld on the nozzle
side at the 12 O’clock position.

4.2 Metallurgy of welds and cracking

Figure 14: Microstructure in the area of typical cracking
In-situ metallography and replication techniques were used next to the outlet nozzle at the 3 O’clock
to determine the structure of the welds and HAZ areas in the position.
cracked regions. Figure 14 shows cracking in the outlet
nozzle that has similar characteristics to the cracking seen in 4.3 Fitness-for-purpose assessment of steam purifier
the brine accumulator with the major difference being that The 22 mm deep flaws in the manway and inlet nozzles
the cracking appeared to have initiated in the coarse grained were selected for fitness-for-purpose assessment.
HAZ where the structure is more prone to hydrogen
embrittlement, however this could not be confirmed as no The fitness-for-purpose assessment was carried out in
laboratory samples were available. accordance with BS7910:1999, Level 2. This method
assesses flaws against both brittle fracture and plastic
8
 Lichti, Firth and Karstensen

collapse. To carry out the assessments Crackwise version 60

3.151 was used, this software follows the procedures 49mm heigh at a stress of
2
127N/mm - likely to cause

Acceptable embedded flaw height,m m

specified in BS7910:1999. This assessment concluded that 50 failure
the longest flaw found in the inlet nozzle was unacceptable,
Figure 15. The maximum allowable flaw depth for a 420 40
Maximum acceptable stress of 100N/mm2
mm long crack was shown to be 18 mm. Although this for avoidance of falilure
30
defect was at the unacceptable limit a more refined
assessments may have indicated acceptability for continued
20
operation but the fully circumferential defect in the manway
was of a similar depth and much longer and was also
10
unacceptable, so the vessel was removed from service for
repair. 0
80 90 100 110 120 130 140 150 160
Prior to repair additional NDT was completed in the region Appied stress, N/mm2
of the reinforcement pads on the inlet and outlet nozzles.
The NDT identified significant regions of lack of fusion,
clusters of small inclusions and small areas of porosity in the Figure 16: Applied stress as a function of acceptable flaw
roots of welds of the external reinforcement pads. Where height for embedded flaws.
practical these were designated for removal but some were
5. CAUSE OF CRACKING
embedded at depths where removal by grinding was not
possible. These impossible to remove defects were assessed The results suggest that the transverse cracking, in the brine
and were considered acceptable under operating conditions accumulator, initiated and propagated along interdendritic
as they fell below the assessment line of Figure 15. boundaries delineated by ferrite in the capping weld. The
cracks were typical of Hydrogen Embrittlement (HE) or
35
Hydrogen Assisted Stress Corrosion Cracking (HSCC).
HSCC is a special case of HE sometimes known as
30 Hydrogen Assisted Cracking (HAC) or more commonly
UNSAFE
Sulfide Stress Corrosion Cracking (SSCC). The cracks may
25
have initiated from corrosion pitting on the capping weld
critical flaw depth, mm

420mm long and 21mm deep (not acceptable)

CRITICAL
20
and the coarse grained aligned dendritic grains produced by
the submerged arc welds appear to be particularly
15 susceptible to corrosion and cracking. The sub-surface
corrosion along the ferrite boundaries also indicates that
SAFE
10 these areas were prone to corrosion. Once the cracking
initiated and a significant stress concentration effect was
5
present the cracking was typical of transgranular HE.
0
0 100 200 300 400 500 600 The degree of corrosion present suggests that the cracks had
Acceptable defect length, mm been present for some time in service. It is not known if the
cracking and corrosion occurred due to HE at a lower
temperature with subsequent high temperature corrosion or
Figure 15: Maximum acceptable flaw length as a due to corrosion and HE at the service temperature. In either
function of flaw depth for inlet/outlet case, the straight nature of the cracking and the presence of
nozzles. multiple parallel cracks indicates that they were primarily
propagating in the presence of a high residual stress.
4.4 Repair and Hydro Test
On completion of the repair a conservative PWHT was During welding high residual stresses are produced that are
designed to give suitable stress relief of the welded areas predominantly parallel with the axis of the weld in
without undue sensitization of the Type 304 stainless steel unrestrained positions. These are often at the yield stress of
lower portion of the vessel and other weld attached internal the material. The stress distribution across a weld depends
fittings. on the sizes of the weld beads and the order of welding.
Where numerous small weld beads are used each weld bead
It was also required to perform a hydro test, however, tempers prior passes and the average stresses are reduced
concern was expressed for the risk of failure from the especially in the middle of double V type welds. However,
embedded flaws under ambient temperature and high where large weld beads are applied, as in this case, the
pressure conditions. converse occurs and high residual stresses are expected
across the entire width of the vessel.
An additional fitness-for-purpose assessment was completed
on the embedded flaws in the reinforcement pads. The It is known that hydrogen cracking can occur through wall in
nominal hoop stress, as a result of a 40.2 barg pressure welds. However, it is more likely that as cracks are formed
hydro test (1.75 x operating pressure as recommended by that there is a balance between an increasing stress intensity
ASME VIII Div 1), was calculated to be 127.5 N/mm2. In due to the crack length and a reduction in the residual stress
accordance to the defect assessment this pressure would due to the presence of the crack and crack opening. As the
cause the 49 mm high flaw that was present to fail, Figure cracks deepen, the later will become dominant and it is
16. The assessment shows that a maximum nominal stress therefore not surprising to see cracks 90% through the wall
of 100 N/mm2 could be tolerated under hydro test thickness which must have effectively stopped propagating.
conditions, this correspond to a pressure of 35 barg. This However, for this size of crack only minor stress increase
pressure was used for the hydro test, being 1.5 x the will cause final failure of the crack ligament.
operating pressure to ensure vessel integrity without risk of
ligament failure in the reinforcement pads. In summary the cause of the cracking was believed to be due
to HE or HSCC in hydrogen charged weld metal in an H2S
9
Lichti, Firth and Karstensen

containing environment. The welding process contributed The concentration of hydrogen in the steel, both in service
significantly to the cracking as the high energy submerged and at shutdown also has a major effect on the susceptibility
arc welds gave a coarse grained microstructure that was to cracking. Hydrogen charging in geothermal environments
more susceptible to cracking than the fine grained is initially high when corrosion first occurs but reduces as
microstructure expected of multi-pass welds. This process protective corrosion products form and the corrosion rate
also leads to high levels of residual tensile stress across the reduces, McAdam et al, 1981. The period of crack
majority of the weld. These remained because the vessel propagation is open to debate. Hydrogen cracking would
was not given a PWTH. normally be expected to be minimal at operating
temperature, rather the hydrogen charged material would
Hydrogen readily diffuses into carbon steel vessel and tend to crack primarily at periods of shutdown when the
pipeline walls exposed to geothermal fluids, McAdam et al, vessel walls were below 100°C, Warren, 1987. However,
1981. Surface hydrogen concentrations are dependant on the corrosion in the cracks occurs at the operating
the corrosion rate, but virtually all of the hydrogen generated temperature and it could be argued that the cracks grow
diffuses into the steel because of the presence of hydrogen when an excessively high level of hydrogen is formed at the
sulfide that acts as a “poison” that promotes uptake of crack tip.
hydrogen. At high temperatures this hydrogen diffuses
through the steel and exits on the outside surface. Exposed The characteristics of the cracking in the two vessels
surfaces can passivate by formation of corrosion products or indicated that the same issues were present in both. They
scales that block the corroding metal from the environment were both made in accordance with ASME VIII Div 1 and
and over time the volume of hydrogen produced and the both had average hardness values below the levels set by
volume of hydrogen in the steel is decreased. The diffusion MR0175/ISO 15156-2:2003(E). However, they were both
of hydrogen out of the metal is temperature dependant and if made using high energy submerged arc welding. This
for example silica scaling occurs at temperatures over 100°C process has become more common over recent years and
and there is rapid blocking of the surface from the corrosive allows deep welds to be produced faster than previously,
solution then the amount of hydrogen generated is likely to thus reducing manufacturing costs. The large weld beads
be small and what is present will diffuse away over a matter produced have a coarse grain structure in the weld and HAZ.
of hours. In addition, the coarse weld beads do not give significant
temper of previous weld passes and as such the residual
The propensity for HE to occur depends on the strength of stresses are potentially at yield throughout the weld.
the steel, the levels of stress, stress concentrations, the
microstructure, composition and hydrogen concentration in The cracking observed here suggests that if cracking is to be
the steel, Lopez et al, 1999. prevented in conditions where high levels of hydrogen can
be present in the steel that additional specifications over
HE is observed in carbon and low alloy steels that have high those required by ASME VIII Div 1 and MR0175/ISO
strength when moderate applied stress are applied. As a 15156-2:2003(E) should be used. For example, to minimize
result of this a common industry standard is to limit the the risk of cracking it is recommended that more and smaller
hardness of these types of steel to 250 HV as in NACE weld passes should be applied and post weld heat treatment
MR0175 and in the new joint standard MR0175/ISO 15156- should be used for vessels approaching the specified
2:2003(E). However, this is NOT an absolute limit and the maximum thickness unless proven otherwise. Methods for
recent joint standard reiterates that the responsibility for proving fabrication variations are outlined in MR0175/ISO
approval of stress levels, heat treatments and hardness 15156-2:2003(E), however, these are difficult to design and
should be with the end user. Testing carried out in prove effective. Our recommendation would be to do
geothermal environments by Marshall and Tombs, 1969, PWHT.
showed that a range of steels suffered from stress cracking in
geothermal condensate when they were loaded to high stress 6. SUBSEQUENT INSPECTIONS
intensities. Even H40 casing steel with tensile strength of
400 MPa (hardness approximately 120 HV) was shown to The replacement brine accumulator vessel was inspected
crack under these conditions, however the applied stresses after one year of service. No cracking was apparent.
required for cracking were above the yield stress. Residual Minimal deposition of silica was noted and only a very thin
stress is known to be sufficient to cause cracking in layer of corrosion products had formed. A small number of
susceptible microstructures, Warren, 1987. shallow pits were found with hollow caps of brown scale
and these were attributed to shutdown corrosion. The pit
Stress concentration has a significant effect on HE. One of caps were removed to allow repassivation on startup.
the major causes of cracking in the steam purifier was the
presence of welding defects which were only identified The repaired steam purifier has been in service for one year
during repair of the vessel. In the brine accumulator the and no problems have been reported.
cracks do go a short distance into the parent metal, again
driven by the existing crack in the weld material. 7. CONCLUSIONS
Numerous transverse weld cracks and a limited number of
The microstructure has a significant effect. Coarse grained longitudinal welds were present in longitudinal seam welds
steels are more prone to HE than fine grained steels and in the brine accumulator vessel after 5 years of service. In
tempered martensitic steels with a fine uniform addition small cracks were seen in the circumferential weld
microstructure are less susceptible than pearlitic steels. areas next to the seam welds.
Continuous bands of ferrite on grain boundaries in welds are
more susceptible to cracking as has been shown in this case. A number of longitudinal cracks were present in the HAZ of
the nozzle welds in the steam purifier.
Minor variations on the composition of low alloy steels have
little effect on the susceptibility to cracking (Ni content is Hardness surveys indicated the vessels complied with the
specifically restricted). requirements of NACE MR0175. The susceptibility of a
steel to Sulfide SCC and HE is dependent on tensile
strength, applied stress, microstructure, composition and the
10
 Lichti, Firth and Karstensen

environment. The hardness limitation of HRC 22 (250 HV) 8. ACKNOWLEDGEMENTS

is normally suitable to help prevent cracking. However, it is
known that steels with hardness significantly lower than this The authors express their thanks to MPT Solutions for
can experience Sulfide SCC and HE if they have susceptible permission to publish this work. SEM support was provided
microstructures and the applied stress is very high ie at or by Kay Card of Industrial Research Limited.
above yield. This was believed to be the case in both vessels
where the cracking observed was typical of Sulfide SCC and 9. REFERENCES
HE and was primarily in the weld material and HAZ but at ASME Section VIII, Division 1 Pressure Vessels: Rules for
times propagated into the surface regions of the parent Construction of Pressure Vessels, Boiler and Pressure
material. Vessel Code, American Society of Mechanical
Engineers, (1997)
ASME VIII Div 1 specifically allows vessels < 32mm wall
not to be stress relieved. Vessels complying with this British Standard Institution, BS7910:1999, “Guide on
standard and MR0175, have for many years given trouble methods for assessing the acceptability of flaws in
free service in vessels containing geothermal environments. metallic structures, 1999.
However, both vessels considered here had significant Lopez, H F, Bharadwaj, R, Albarran, J L and Martinez, L,
cracking present that was not fit for long term service. The Role of Heat Treating on the Sour Gas Resistance
of an X-80 Steel for Oil and Gas Transport, Met and
The large submerged arc weld passes used produced a
materials Trans A, Vol 30A, Sept (1999), 2419-2428.
coarse grained microstructure in the weld and a coarse
grained HAZ. This type of welding has become McAdam, G D, Lichti, K A and Soylemezoglu, S Hydrogen
commonplace for fabrication of ASME VIII Div 1 vessels. in Steel Exposed to Geothermal Flyuids, Geothermics,
However, the process results in high residual stress and an Vol 10, No 2, (1981), 115-131.
unacceptable microstructure for resistance to Sulfide SCC
and HE in geothermal environments. Marshall, T and Tombs, A, Delayed Fracture of Geothermal
Bore Casing Steels, Australasian Corrosion
The replacement brine accumulator vessel, fabricated using Engineering, Vol 13, No 9, September (1969), 1-8.
conventional multi-pass welding with a PWHT was found to Milliams, D E and Tuttle, R N, ISO 15156/NACE MR0175
be crack free after one year of operation. – A New International Standard for Metallic Materials
for Use in Oil and Gas Production in Sour
The repaired purifier vessel has been in service for one year
Environments, Stainless Steel World, September,
with no known problems.
(2003), 63-68.
Cause of Cracking NACE MR0175 (Periodic to Annual Revisions to 2003)
The main cause of failure was the use of submerged arc Standard Recommended Practice for Sulfide Stress
welding with only 10 weld passes and a lack of PWHT on Cracking Resistant Metallic Materials for Oilfield
the 32 mm wall thickness welded vessels operated in an H2S Equipment, NACE International, (2003).
environment. The cracking in both vessels was attributed to NACE MR0175/ISO 15156-1:2001(E) Petroleum and
SSCC and HE. The welds were particularly prone to this as natural gas industries---Materials for use in H2S-
they had a coarse structure, a coarse HAZ and high residual containing Environments in oil and gas production---
stress throughout. Part 1: General principles for selection of cracking-
resistant materials, First edition, NACE
Avoidance In Future International/ISO.
ASME VIII Div 1 allows vessels of < 32 mm wall thickness NACE MR0175/ISO 15156-2:2003(E) Petroleum and
not to be heat treated. The requirement is noted in NACE natural gas industries---Materials for use in H2S-
documents to be marginal and suggests if a PWHT is to be containing Environments in oil and gas production---
avoided then testing of heavy walled fabricated components Part 2: Cracking-resistant carbon and low alloy steels,
is required to agreed conditions. However these are difficult and the use of cast irons, First edition, 2003-12-15,
to do and can be costly. Alternatively all vessels NACE International/ISO 2003.
approaching this thickness should be given a PWHT.
NACE MR0175/ISO 15156-3:2003(E) Petroleum and
In our opinion, avoidance of HE cracking in vessels used for natural gas industries---Materials for use in H2S-
geothermal service requires an additional set of “rules of containing Environments in oil and gas production---
thumb”: Part 3: Cracking-resistant CRAs (corrosion-resistant-
alloys) and other alloys, First edition, 2003-12-15,
1. The number of weld passes must be as many as the wall NACE International/ISO 2003.
thickness in millimeters i.e. for a 32 mm wall at least
32 passes should be applied. TWI software program, Crackwise v. 3.15, BS7910:1999
2. All vessels should be stress relieved unless it can be fracture/fatigue assessment procedures.
proven that this is not necessary by the manufacturer Warren, D, Hydrogen Effects on Steel, Materials
for the welding procedure chosen. performance, NACE, Jan, (1987), 38-48.
3. In addition, care should be taken to ensure any closing
welds are correctly stress relieved.

11