Paper 5c

Primary Reformer Tubes Failure Due

to Corrosion Attack
For the last 30 years reformer tubes have gone through numerous changes in manufacturing
processes and material of construction to improve performance and reliability. To increase plant
throughput QAFCO replaced the old HK40 tubes with HP modified alloy but within twenty-four
months of operation, unexpected corrosion attack was observed on some of the reformer tubes. In this
paper the potential root cause of the tube damage and remediation actions has been discussed.

Iftikhar Turi and Shailender K Bhandari

Qatar Fertilizer Company, QATAR

Introduction located on either side of the reformer tubes facing

downward.

Q
atar Fertilizer Company (QAFCO)
located in Qatar (Middle East) is one of In 2003 ammonia-2 primary reformer was re-
the leading producer of urea and ammonia tubed by replacing HK-40 tubes with Nb-
in the world. At present, QAFCO operates four modified HP alloy having the same outer
ammonia and four urea plants. The present diameter but reduced wall thickness. In 2005
annually production is 2.2 MMT of ammonia and ammonia-1 reformer also underwent the same
3.0 MMT of Urea. change.
QAFCO is in expansion process, to increase its After about twenty four months of the plant
ammonia and urea production by installing two startup, during routine visual inspection plant
ammonia (2300MTPD each) and one urea operator observed unusual spots on some of the
(3850MTBD) plant. The installation and reformer tubes. Initially, the spots were
commissioning is expected to complete in 2011. considered to be a salt deposition or insulation
Both QAFCO 1&2 plants are Davy Power design material sticking to the tube wall and glow but
and were commissioned in 1973 and 1979 during November 2007 turnaround, severe
respectively. Both the plants are identical to each degradation and loss of material was observed at
other, has a top fired reformers containing 300 isolated locations on several tubes. Out of 300
tubes arranged in 10 rows of 30 tubes, with a tubes, 65 tubes were replaced as their wall
heated length of 40 feet. The burners are arranged thickness was reduced significantly. The same
in eleven rows consisting six burners in each row, effect was expected on the ammonia-1reformer
tubes as well because both the reformer has the

2009 255 AMMONIA TECHNICAL MANUAL

 same kind of tubes and share the same quality of increase plant capacity. The reformer capacity
fuel gas. (Fig 01, 02) was up-rated by raising the operating temperature
During 2008 ammonia-1 turnaround the same by 20-30°C. The temperature rise was achieved
phenomenon was observed on several tubes. To by increasing fuel to reformer burners without
avoid premature failure all the affected tubes with changing or modifying 20 years old burners.
reduced wall thickness were replaced. After about 24 months of successful startup and
In order to monitor the corrosion activity closely, achieving the expected increase in plant load,
frequent thermography and visual inspection unusual spots started to appear on some the
were conducted and possible control measures reformer tubes. Most of the spots were located on
were taken to slow down the tube degradation the lower part of the tubes, while some tubes
process. Due to this unexpected tube damage, were attacked over the entire heated length. In
investigation was launched to find the root cause. some cases the adjacent tubes were found
Both Yara International (Norway) and Quest unaffected. The corrosion products found on
Reliability (Johnson Matthey) were requested to these tubes were thick and voluminous caused by
help in this investigation. the loss of parent material. (Fig 03) The spots were
round in shape forming a thick crust with raised
surface varying from small nucleation to large
areas. Extensive loss of material was also
observed from the bigger spots. (Fig 01, 02)
The affected tubes profile created by routine
observation of spots on individual tubes indicates
that the majority of the affected tubes are on the
south side of the reformer. (Fig 04)

Fig # 01ammonia-2 reformer

Fig # 03 ammonia-2 reformer

Fig # 02 Ammonia-2 reformer

Fig # 04 ammonia-2 reformer affected tubes
Ammonia - 2 Reformer Tubes Damage profile
As mentioned earlier, ammonia-2 plant, original
HK40 reformer tubes were replaced after 24
years with Nb-modified HP alloy in 2003 to

AMMONIA TECHNICAL MANUAL 256 2009

 Findings in Ammonia - 1 Reformer levels in the firebox. The north side of the
Both ammonia 1&2 reformers are identical and reformer had the majority of the high readings
share the same quality of natural gas. As indicating that this zone of the primary reformer
mentioned earlier, ammonia-1 reformer was also has insufficient combustion air. Some relatively
re-tubed in 2005 by replacing old HK40 with the high readings were observed on the south side of
Nb-modified HP alloy tubes. On November 14, the reformer also.
2007 a thorough inspection of the ammonia-1
Oxygen Survey
reformer tubes was conducted from the lower
peepholes. Around 20 tubes were found to have Significant excess air in the south side of the
the same kind of attack as observed on ammonia- primary reformer but much lower levels in the
2 reformer tubes. During November 2008 north side of the primary reformer outer rows of
turnaround, around 80 tubes were found to have burners have in general a higher oxygen level
corrosion attack (Fig 05) than the center rows. The variation in oxygen
levels is consistent with the variation observed in
CO levels.

Temperature Mapping
The corrected tube wall temperature for the top
and bottom portion of the reformer tubes
indicates that the south side of the reformer is
hotter than the north side and the tube damage
tends to fall in the hot areas of the primary
reformer. (Fig 06)
Fig # 05 ammonia -1 reformer

Combustion Survey
Visual Inspection of Flame Characteristics
During visual inspection of the firebox conducted
by Hamworthy Combustion Engineering Limited
many of the burners flame were observed to be
unstable rather long straight and narrow and were
moving towards the reformer tubes. A number of
burners were observed to have the flame slowly
moving away from the burner because of
irregular air supply probably due to partial
blockage of nozzles.
Some haziness was also observed at the lower
level indicating after burning due to insufficient
combustion air in some parts of the primary
reformer.

Carbon Monoxide Survey

During the CO survey conducted from peepholes
at different levels close to the reformer walls, it Fig #06
was found that there is wide variation in CO

2009 257 AMMONIA TECHNICAL MANUAL

 Tube Microstructure Metallurgical The specimens prepared were first examined in
Examination and Corrosion Product the unetched condition and selected features were
chemically analyzed by using (SEM, EDX). The
Analysis
same specimens were the etched and again
Visual Inspection examined under optical microscope.
The corrosion attack on the outer tube surface are
varying from small nucleation points to large
areas with extensive loss of wall thickness.
Optical microscopy shows that the corrosion
attack was transition type, between type 1 & II of
sulfidation, deep penetration with finger like
projections and uneven scale-metal interface
containing sub-surface sulfides. (Fig #07)

Fig #08 ammonia -2 reformer tube samples

Sample B
A relatively high level of aluminum was detected
Fig #07 in the concave surface of the loose piece of scale,
presumed to have been adjacent the surface of the
The depth of penetration of the localized tube. This was probably present as Al2O3.
corrosion product was 0.25 mm on the outside Sulfur was present to high levels in the majority
surface. of the corrosion product (20911c Table 02).
Corrosion product at the corrosion front was Corrosion product with the highest levels of
associated more with the white chromium-rich sulfur (20911c Table 02) also contained the highest
primary carbides than with the Nb-rich carbides. level of Fe and almost the lowest level of Ni.
Small spots of corrosion were present at TiN Corrosion product at the extremity of the finger-
precipitates. like grain boundary attack (20911l Table 02) and
(20911m Table 02) contained considerable levels of
Chemical Analysis / Microscopic Examination O and little S.
High levels of Mn and Cr were associated with
Three samples were prepared and marked
high O in the finger-like corrosion projections.
accordingly for comparison: (Fig 08)
Sulfur was not detected at high levels in
B - an affected bottom portion of the tube.
association with primary carbides containing
2 - an unaffected portion of the tube from
appreciable Nb.
about two meters above B.
High O levels were present in material very near
Q - top portion of the tube, about 5 m from the
the surface and at the finger-like extremities of
very top of the tube.
corrosion attack and were associated with high
Loose corrosion product.
levels of Cr.
A sample of material was analyzed by using OES
Na, K and Mg were not detected in the
and EDX method for comparison. (Table 01, 02)
metallographic section and Ca and V were

AMMONIA TECHNICAL MANUAL 258 2009

 present at low levels in only two analyses.
However, both Ca and V were present in analysis
(20991f Table 02), which was taken at the finger-
like projection in the corrosion front.
Sulfur was not detected in the secondary carbides
within the highly populated near- surface region.
Chromium was the predominant element. These
were most likely to be M23C6 chromium Spectrum 20911O
carbides.
Sulfur was not detected in the matrix surrounding
the above mentioned chromium carbides.

Spectrum 20911P

Sample Q
Corrosion morphology and severity in sample Q
was very similar to sample 2. No sulfur was
detected in the corrosion product analyzed on this
sample.

Loose Corrosion Product

The corrosion products found on the surface of
the tube were found to be thick and voluminous.
The analysis of the parent material indicates the
Sample 2 loss of parent metal to form the corrosion
products.
A TiN precipitate cluster near the surface that The convex side of the large loose lump
was associated with one of the largest corrosion contained major amount of magnetite (Fe3O4),
penetrations, was analyzed, represented by minor hematite (Fe2O3), heazlewoodite (Ni3S2)
spectrum (20911O) and (20911P). Other elements and a minor/trace amount of bunsenite (NiO).
found to be associated with the precipitate
included Al, O, Nb and Ca. The aluminum may
have been present as alumina, or Al2O3. The
corrosion products in the same penetration shown
in fig #10 include probably Al2O3 while the
surface product contains Si, Fe, Mn and Cr.
No sulfur was detected in any of the corrosion
products analysed on this sample, indicating
normal oxidation in the presence of TiN.
(Fig #09)

The concave side contained a major amount of

magnetite-like spinel (FeCr2O2), minor eskolaite

2009 259 AMMONIA TECHNICAL MANUAL

 (Cr2O3), and a minor amount of heazlewoodite, microstructure of the base material is according
metal, and may be CrNbO. to the specified alloy composition. (Table #01)
The presence of nickel and iron oxide in the
presumed outer regions and chromium and iron- Lab #1
chlorides closer to the tube’s surface in the Element (%) Ni Cr Fe Nb
detached lump of corrosion was typical of Base material 34.9 25.4 36.1 1.0
corrosion product resulting from carburization HP Nb-modified 35 25 37 1
and sulfidation of this alloy. The presence of Lab #2
nickel sulfides in this corrosion product is Element (%) Ni Cr Fe Nb
consistent with sulfidation. Base material 34.32 26.27 35.6 0.77
Corrosion products in the same penetration, HP Nb-modified 35 25 37 1
shown in its entirety is included probably Al2O3 Table 01
and SiO2 in one finger-like projection (20911p)
and some oxides of chromium in another finger-
like projection, and predominantly Al2O3 in a Key findings and observations
third finger-like projection. (Fig 10) A combination of the burner geometry anomalies,
over firing and insufficient or poor combustion
air flow distribution created localised sub-
stoechiometric (reducing) conditions within the
radiant section of the primary reformer, resulting
in poor combustion, irregular flame patterns and
high levels of CO are indicative of protective
oxide layer damage caused by surface pitting and
scale formation. Reducing conditions such as
these are known to be detrimental to the process
tubes, often causing scaling and surface pitting
removing or damaging the protective oxide skin.
Fig #10
The corrosion products formed are consistent
The chemical composition of the corrosion with sulphurdising with hydrogen sulphide in the
products varied. The bulk of the product just fuel gas of 0.6% and high partial pressures of S2
below the surface contained mostly Al and O, of 10-8 atm at high temperature. Increasing
probably as Al2O3 while the surface product, also temperatures leads to a higher rate of
contained oxides of Si, Fe, Mn and Cr. sulphidation. The mixed oxides and sulphides
The presence of nickel and iron oxides in the present in the corrosion products indicate that the
presumed outer regions and chromium and iron- flue gas around the affected tubes was not
chromium oxides closer to the tube’s surface in constant and was alternating between oxidising
the detached lump of scale was typical of and sulphurdising.
corrosion product resulting from carburisation High temperature sulphidation can be made
and sulfidation of this alloy. The presence of worse if there is hydrogen present, as the rate of
nickel sulfides in this corrosion product is corrosion is directly related to the hydrogen
consistent with sulfidation. (Fig 01&02) sulphide partial pressure. At high temperatures
(greater than 500°C), as the hydrogen sulphide
Chemical Composition partial pressure rises the rate of corrosion
Chemical composition of the tube base material increases dramatically.
conducted in two different labs indicates that the Considerable amount of sulfur is detected in
sample B. However, no sulfur is detected in

AMMONIA TECHNICAL MANUAL 260 2009

 samples 2 and Q which exhibited minor corrosion sulfide, which is non-protective, causing alloy
typical of high temperature oxidation. exposed to corrosion attack.
The morphology of the severe corrosive attack on 3. When the environment is oxidizing the alloy
sample B is consistent with sulfidation as is more likely to form a protective chromium
heazlewoodite (Ni3S2), has been detected in the oxide scale, rather than a chromium sulfide.
scale examined by using XRD. 4. As the corrosion attack happened mostly on
A high population of secondary carbides has been the lower portion of the tubes, shows that the
detected beneath the sulfidation in the near furnace environment is changing from being
surface material, indicating that carburisation had oxidizing to reducing due to oxygen depletion
probably occurred. However, carburisation was with downward flow of fuel gases in the fire
not present in material adjacent the sulfidised box. Metal-metal sulfide eutectic i-e nickel-
regions. This indicates that carburisation and nickel sulfide (Ni- Ni3S2), magnetite (Fe3O4)
sulfidation only occurred together and not etc were also detected in the corrosion
independently. product.
The microstructure of the bulk material well 5. In the reducing environment, when the sulfur
beneath the surface structure was similar in all is too high, the alloy will form a chromium
samples, i.e. sample B did not exhibit a sulfide rather than an oxide. While the oxide
microstructure that could be considered to be may be protective, the sulfide offers little
more susceptible to corrosion than the other resistance to corrosion attack. Nickel reacts
samples, except in the surface carburised regions chemically with sulfur very readily forming
beneath the corroded areas. The similar bulk metal sulfides. Metal sulfide or metal-metal
material microstructures of the three samples sulfide eutectics (Ni- Ni3S2) are often molten
indicated that the temperature in all three regions (635°C) at operating temperature of the
of the tube from which the samples were taken primary reformer.
was fairly similar. This indicates that region B 6. Corrosion damage of sample B probably
was probably no more susceptible to corrosion included the formation of liquid eutectic
attack than regions 2 and Q, based on temperature nickel-nickel sulfide (Ni3S2).
only. This suggests that the sulfidation in sample 7. The mixture of sulphides and oxides in the
B was probably driven more by physical and corrosion products indicates that the flue gas
chemical aspects than by the temperature of local to the corroded tubes was alternating
exposure. between an oxidising and reducing
atmosphere due to imbalance reformer.
Conclusion 8. To withstand the extreme operating
1. Microscopic examination, chemical analysis, conditions of temperature and pressure,
combustion survey and the high sulfur content micro-alloys have made it possible to increase
in the corrosion products reveals that the the capacity of a plant by maintaining the tube
reformer tubes failure occurred due to outside diameter and decreasing the wall
Sulfidation of a transitional type between thickness but somehow made it vulnerable to
Types I and II, preceded by carburization. sulfidation attack due to increase in nickel
2. When the combustion of the gas containing contents.
sulfur takes place under stoichiometric and 9. Prior to reformer re-tubing QAFCO was not
sub-stoichiometric conditions, sulphur is cautioned about the HP micro alloy tubes
converted to H2S and the environment susceptibility under reducing environment to
becomes reducing. Under reducing sulfidation attack due to the presence of sulfur
environments the alloy forms chromium in the fuel gas.

2009 261 AMMONIA TECHNICAL MANUAL

 Action Taken air ingress to get the representative sample of
With change in fuel gas quality and increasing the the flue gas.
combustion air:
Observations after actions taken
1. Reduce Hydrogen Sulfide contents in the fuel With change in fuel gas quality and increasing the
gas to reformer. combustion air, after six months of service no
Both Yara, Quest Reliability and Literature corrosion attack has been observed on ammonia-1
Survey indicates that, under reducing reformer tubes.
conditions in the fire box fuel gas containing
more that 0.6% Hydrogen Sulfide can lead to And the appearance of new corrosion spots on
sulfidation attack on the (Nb-modified HP ammonia-2 reformer tubes are on the decreasing
alloy) reformer tubes. trend.
A modification was conducted to pass the
fuels gas through sweetening plant before Acknowledgement:
mixing with purge gas. It reduced the This paper is prepared under guidance and support of
Hydrogen Sulfide contents up to 0.02% in the QAFCO Management and the following individuals and
departments also contributed in the study and compilation
fuel gas. of this paper.
2. The combustion survey conducted on the Departments: Inspection & Material Technology,
primary reformer indicates that there is an Field Maintenance, Laboratory and f ammonia1 & 2
issue with the combustion air distribution Operation Staffs.
across the reformer creating localized sub- Individuals:
Richard de la Bastide, Head of Ammonia-3 and acting
stoichiometric conditions due to insufficient Head of Production Support, Mr. Geoffrey Blewett, Head
combustion air in some parts of the primary of ammonia-1 and Mr. Rob Stevens ex- Head of ammonia-
reformer. 3 and acting Head of Production Support.
To improve the air distribution the burners
combustion air dampers and fuel valve were References
fully opened. 1 QAFCO Inspection & material Technology, “Primary
Burners wind box pressure survey was Reformer Tube Skin Temperature Measurement Report.”
conducted to ensure equal distribution of 2 Johnson Matthey (Quest Reliability), “Primary Reformer
Tube Corrosion Damage Assessment for QAFCO.” Rev
combustion airflow to the individual burner. 1.
Frequent reformer tubes temperature mapping 3 Hamworthy Combustion Engineering Limited, “
and visual inspection is being conducted to Combustion Survey on the Ammonia -1 and Ammonia –II
detect any abnormality with in the fire box. Reformers.”
Excess O2 in the furnace discharge has been 4 Yara International ASA, “QAFCO-2 Primary Reformer
Tube Failure.”
increased from 0.8% to ~1.8 to 2.0 % in both
ammonia plants.
Air ingress check through convection zone
was conducted by smoke test to eliminate the

AMMONIA TECHNICAL MANUAL 262 2009