Lessons Learned from High Pressure

Process Boiler Failure

After 13 months of service, a new 1500 psig (103 barg) secondary reformer effluent process boiler de-
veloped a severe leak. After stabilizing the plant, operators identified the nature of the issue. Precau-
tions were taken during the shutdown that were successful in preventing catalyst damage in the sec-
ondary reformer and shift converters due to the leak of BFW into the process gas. The tube bundle
was removed and replaced with a spare. The immediate cause of the failure was identified and a met-
allurgical analysis was performed on the failed and adjoining tubes. Possible contributing factors
were considered. Metallurgical analysis results are consistent with the cause identified. Recommen-
dations are shared on promptly recognizing this type of failure, shut down precautions, and preven-
tion.

Larry Walker
Project Manager

Keith Wilson
Ammonia Technology Manager

PCS Nitrogen Fertilizer LP

Augusta, Georgia, U.S.A

eration. The process effluent feeds the high tem-

perature shift converter (HTS).
Introduction

P CS Nitrogen Fertilizer LP operates a 2425

stpd (2200 mtpd) ammonia plant at Au-
gusta, Georgia, USA. The plant was start-
ed up in 1978 with a capacity of 1500 stpd (1360
mtpd) and has been debottlenecked in several
Process Boiler Installation
The process boiler uses forced-circulation of wa-
ter on the tube side, and a rod baffle design to
minimize the process pressure drop on the shell
steps. During a maintenance and expansion side. Please refer to figure 1. The tube count is
turnaround in 2012, the 1500 psig (103 barg) 708 one inch U tubes and the design duty is 336
process waste heat boiler was replaced with a MMBtu/hr (98.5 MW). Design vaporization is
new larger boiler designed by KBR. This boiler 20% by weight of the circulating water. The ex-
cools secondary reformer effluent by steam gen- changer has a removable bundle of U tubes and

2014 267 AMMONIA TECHNICAL MANUAL

 is mounted vertically with the tube sheet at the
top. Shell side flow is single pass. The tubes are Start up was normal and the exchanger achieved
Inconel 625 based on our very good experience the expected performance for heat duty and pres-
with its resistance to metal dusting in this ser- sure drop.
vice. The operating steam to carbon ratio is 3.0
mol/mol. The exchanger has internal refractory Failure symptoms
lining on the shell side.
On November 24, 2013 the following symptoms
1500 psig / 103 bar g
Saturated Steam were observed by the operations team:
1) Sudden 20% drop in high pressure steam
Boiler Steam Drum
generation. This caused loss of speed on
Feed
Water
598°F/314°C the syngas turbine and high syngas com-
pressor suction pressure.
2) Drop in high temperature shift converter
From
inlet temperature from 662 F (350 C) to
Synloop
Boilers
605 F (318 C).
To HTS
3) High level in process condensate separa-
Converter
662°F/350°C
tor upstream of the CO2 absorber
4) Initial drop in level in the CO2 stripper,
To
Synloop
after which the level lined out
Boilers From Secondary
Reformer
5) Drop in phosphate level in the continuous
Boiler Circulation 1550°F/843°C high pressure steam drum blow-down
Pumps (3) U-tube
Process Boiler 6) Presence of phosphate in a process con-
Figure 1 - sketch of forced-circulation boiler flow densate sample taken upstream of the
circuit HTS vessel.
To accommodate higher boiler circulation water The drop in high pressure steam generation was
flow and higher steam generation, boiler inlet caused by the quenching effect of a major boiler
and outlet piping was modified. Due to reaching circulation water leak into the process. Based on
end of life, new boiler feed water preheat coils a heat balance, the BFW leak into the process
were installed in the reformer convection section. was 122,000 lb/hr (55.3 mt/hr)! The leak also
Steam drum nozzles and internals were also caused the drop in the HTS inlet temperature of
modified during the turnaround when the new 57 F (32 C). The high level in the process con-
boiler was installed. densate separator was also caused by this addi-
tional water. The leaked water was vaporized
During installation, careful attention was paid to before the HTS but was then condensed in the
cleanliness, with Operations conducting the final CO2 reboilers. This extra condensate was sepa-
cleanliness inspection prior to closing of equip- rated in the process condensate separator and
ment. A flushing plan was executed to remove overloaded the process condensate pump, result-
any debris from piping prior to closing equip- ing in high level.
ment. The boiler circulation pumps could not be
operated during flushing, so the flushing velocity The condensation of this water in the CO2 re-
was lower than the normal velocity during opera- boilers increased stripping steam and caused the
tion. Flushing was performed with demineral- initial loss of solution level in the CO2 stripper.
ized boiler feed water (BFW). There was no The level became so high in the process conden-
chemical cleaning of the exchanger or BFW cir- sate separator that condensate carryover devel-
cuit, but the exchanger tubes were pigged in the oped into the CO2 absorber. This led to the sta-
shop before shipment.

AMMONIA TECHNICAL MANUAL 268 2014

 bilization of the solution inventory in the system Normally during shutdown the process is fully
as shown by the CO2 stripper level lining out. depressured and purged three times with nitro-
The condensate carry over offset the additional gen. This removes process steam and flammable
boiling and the stripper level stabilized. gases, after which the process can be placed un-
der low nitrogen pressure. This procedure was
The high pressure boiler chemistry is monitored modified to pressure up just once with nitrogen
by analyzers on the continuous blow-down from and hold the nitrogen pressure until the steam
the steam drum. The analyzer showed low phos- system pressure was lower than the process (ni-
phate indicative of excess blow-down which was trogen) pressure. This stopped the leakage as
occurring through the leak into the process. The soon as possible.
operators pulled a sample of process gas up-
stream of the HTS vessel, condensed the process These precautions were successful in preventing
steam in the sample, and tested for phosphate. damage to the secondary or HTS catalyst. Au-
The presence of phosphate indicated a leak of gusta’s HTS converter is a radial flow design
boiler circulation water into the process. from Casale SA. The radial flow design, with its
large flow surface area, is excellent protection
The operations team was able to stabilize the from solids build-up due to boiler leaks. HTS
plant and steam system, and plan an orderly pressure drop was unchanged as a result of the
shutdown to repair the major leak. tube failure.

Shutdown precautions Bundle removal and first inspection

It was essential to prevent the leaking boiler cir-
On removal of the bundle, a ruptured tube was
culation from backflowing into the secondary re-
found. Some minor damage was also visible to a
former or carrying forward into the HTS vessel.
few tie rods and rod baffle attachment welds.
We also wanted to minimize wetting of the re-
This was likely due to thermal stresses in the
fractory lining in the process boiler and second-
bundle caused by the large leak and local
ary reformer. Although HTS catalyst can toler-
quenching effect.
ate some wetting, our experience was that hot
secondary reformer catalyst is damaged when
quenched with water. The result is higher pres-
sure drop on the secondary reformer. The follow-
ing precautions were taken to prevent the leaking
boiler circulation water from reaching either the
secondary reformer catalyst or HTS catalyst dur-
ing the shutdown:
1) Prompt opening of all available drains
around the secondary reformer, process
boiler, and HTS vessel
2) Prompt pressure reduction of the 1500
psig (103 barg) steam system to reduce
inflow of boiler water through the leak in- Figure 2 - tube bundle removed for repair
to the process side of the boiler
3) Modification of the normal post shut- Several pieces of debris were found on the inlet
down process purging procedure. tube sheet. In addition to some spiral wound
Once the steam system pressure was equalized gasket media and some loose weld slag, we also
with the process pressure, the leak would stop. found two thin carbon steel strips. One was 3” x

2014 269 AMMONIA TECHNICAL MANUAL

 1” x 3/16” (75 mm x 25 mm x 5 mm) and the
other was 1-1/2” x 1” x 3/16” (38 mm x 25 mm x We were fortunate to have a spare bundle on site.
5 mm). Each of the metal strips showed some If we did not have a spare, we would have con-
wear. The longer strip showed two faint but dis- ducted a field eddy current inspection of each
tinct wear circles the same diameter as the tubes. tube. The damaged tubes would have been iden-
Two half circles were faintly visible on the tified, plugged, and the exchanger returned to
smaller strip. It was apparent that these strips service. The duration of the outage would have
were sitting on the inlet tube sheet and blocking been longer.
boiler circulation flow into the tubes on which
they rested. The strips must have sat there for There was potential for water exposure to several
some length of time for the wear patterns to de- process areas: Secondary reformer catalyst and
velop. The tube to tubesheet welds create slight refractory lining, the refractory lining of the pro-
ridges on the tube sheet at the inlets of the tubes. cess boiler, the HTS catalyst and potentially even
These ridges had worn slightly into the metal the low temperature shift (LTS) catalyst. The
strips. The wear patterns can be discerned in the leak would have created extra process steam and
photograph. raised the LTS vessel inlet dew point, possibly
resulting in wetting of this catalyst as well.

We followed vendor’s recommendations by con-

ducting a dry out procedure as part of the start
up. Nitrogen was circulated using the feed gas
compressor. This was accompanied by gradual
heating and holding as water was recovered in
the suction separator for the feed gas compressor.

After start up, we checked for the presence of

Figure 3 - debris found on tube sheet showing
any new hot spots on the shell of the process
imprint of tubes boiler – none were found. This indicates the re-
fractory was not significantly affected. Catalyst
Since the steam drum internals were modified pressure drops also appear to have been unaffect-
during the turnaround when the new boiler was ed.
installed, we carefully reviewed the drawings for
any components of the approximate size of the
two metal strips found. There were no matches,
Bundle inspection and repair
so the source of the metal strips is unknown at The bundle was transported to the vendor’s shop
present. During the next turnaround, we will in- for thorough inspection and repair. An eddy cur-
spect the steam drum internals for any missing or rent technique was used to inspect each U-tube.
loose pieces. This inspection found one bulged tube adjacent
to the failed tube. No other tubes showed any
Precautions taken during restart damage or defects. Finding a damaged tube ad-
jacent to the failed tube is consistent with the
The plant was down for five and a half days to wear pattern on the longer strip. It appears the
remove the damaged tube bundle, inspect the in- strip was blocking these two tubes, resulting in
ternal refractory lining, install the spare bundle, damage to both and eventual failure of one.
and restart the plant. The internal refractory lin-
ing was in excellent shape and withstood the Each U-tube was also pigged in the shop to en-
quench water flow well. sure that there was no debris lodged in any tube.

AMMONIA TECHNICAL MANUAL 270 2014

 Unusual failure mechanism
PCS Nitrogen’s representatives worked in con-
cert with specialists from KBR to develop a plan At first glance the failed tube appeared to have
to repair the minor tie rod and rod baffle weld suffered a fish mouth rupture very typical of se-
damage. PCS Nitrogen elected to replace the vere overheating. Closer examination revealed
two damage tubes. This would require removing that substantial thinning had occurred prior to the
and replacing 44 tubes to gain access to the dam- failure, and that the failure itself was brittle in
aged tubes. All repairs were performed in the nature. IMR reported that heavy deposits were
vendor’s shop. found in the failed area of the failed tube. De-
posit depth in sound tubes was minimal and as
expected for this service. IMR reported that cor-
rosion associated with the deposits caused the
Metallurgical failure analysis tube thinning which led to the tube failure. This
is supported by our own stress calculations based
Samples of the failed tube and some tube sec- on the observed thickness at the failure site and
tions from undamaged tubes were submitted for expected temperature.
metallurgical analysis. The samples of undam-
aged tubes came from tubes that were removed
to gain access to the damaged tubes.

Figure 4 - failed 1" (25 mm) tube

Figure 5 - section of failed 1" (25 mm) tube show-
The metallurgical analysis was performed by ing thinning prior to brittle failure
IMR Metallurgical Services of Louisville, Ken-
tucky, USA. The chemical composition and av-
erage wall thickness were consistent with the de-
sign specifications for the tubes, specifically The blockage at the tube inlet must have been
Inconel 625 Grade 1 annealed tubing with 0.080 partial, and not complete. Deposits were formed
inch (1.27 mm) average wall thickness. Tensile where the combination of low circulation water
strength was tested and found to be good. Tube flow and high heat flux produced higher than in-
hardness was slightly higher than expected, but a tended vaporization. Corrosion associated with
ductility test showed the ductility was adequate the deposits gradually thinned the tube until the
for the bending needed to make the U-tubes. brittle failure occurred.
IMR concluded there were no fabrication or ma-
terial deficiencies in the tube samples.
Lessons Learned
We suggest that operations teams review togeth-
er the symptoms of secondary reformer effluent

2014 271 AMMONIA TECHNICAL MANUAL

 process boiler leaks, so they can be promptly Disclaimer
recognized and a quick but orderly shutdown can
be executed. Not all leaks will be as severe as Potash Corporation of Saskatchewan, Inc and its
the one described here, but when several of the direct and indirect subsidiaries (PotashCorp) are
six symptoms listed above are found, a leak is not responsible and assumes no obligation or lia-
indicated. bility, in contract, in tort, in negligence, or oth-
erwise, for the information provided by the au-
Operations teams can also review and plan the thor, including, without limitation, opinions,
precautions to be taken during shutdowns when a judgments or statements contained herein. The
leak is evident or suspected. This will help avoid information reflects the views of the author, is
damage to refractory or catalyst. Occasional subject to change and does not necessarily reflect
checking of drains is also needed since they have the views of PotashCorp. PotashCorp makes no
a tendency to plug with loose refractory or cata- representation or warranty, express or implied, as
lyst dust. With appropriate safety precautions, to the accuracy, completeness, usefulness, results
plugged drains can be cleared with the plant on obtained from the use of this information or any
line. These process drains will be essential dur- other matter related to the information contained
ing shutdown should a leak occur. herein. In no event shall PotashCorp, the direc-
tors, officers, employees or agents thereof be lia-
After any piping or convection coil modifica- ble for any decision made or action taken in reli-
tions, we suggest using temporary strainers in the ance on the information contained herein or for
pump suction of forced-circulation boiler sys- any direct, indirect, incidental, special or conse-
tems even when careful flushing has been quential damages.
planned and executed. It is seldom possible to
flush with water velocities as high as normal op-
eration, so debris not removed during flushing
may find its way to the inlet tubesheet area and
cause boiler damage due to restricted flow. After
our leak, as a precaution, we elected to install
temporary strainers in our circulation pump suc-
tions. These will be removed next turnaround if
they are found clean.

It is also possible to include a permanent debris

strainer or basket in the design of a process boil-
er. Such a strainer or basket would be located in
the exchanger channel just prior to the inlet tube
sheet. Its advantage is its permanent location and
the fact that it is downstream of all piping, and
can catch any debris, including debris originating
downstream of the circulation pump.

In a new plant design, a forced-circulation boiler

could have a vertical orientation with the chan-
nel/tube sheet on the bottom and the U tube
bends at the top. Any loose debris that entered
the channel would sink to the bottom, and not
block any tubes.

AMMONIA TECHNICAL MANUAL 272 2014