Failure of LP Flash Drum Packing Bed Support

The QAFCO-3 ammonia plant was tripped due to power failure. During start-up of CO2 removal
section, semilean solution pump tripped on low flow. Both semilean and stripper feed pumps strainer
found choked with 2” CS pall ring packing. LP flash drum bottom manhole opened subsequently and
found around 50 m3 (1766 ft3) of packing in sump area. Packed bed inspected and two bottom plates
of packed bed support found lifted from shell support ring towards upside direction by vertical force.
Packing from LP flash bottom sump unloaded and reloaded back in the column. Bottom packing bed
support repaired and production resumed after 9 days shutdown. The paper describes the root cause
of packed bed failure, bed support repairs and post repairs experience.

Geoff Blewett
Qatar Fertiliser Company, QATAR

Umesh R Desai
Qatar Fertiliser Company, QATAR

Introduction
Rich amine solution is regenerated in a two

Q atar Fertilizer Company S.A.Q. (QAFCO)

is located in Mesaieed, Qatar and operates
six ammonia and five urea production
stage flash drum and a stripper.

The CO2-enriched solution from the bottom of

trains. Sixth urea train is under construction. the absorber at 3350 kPag (486 psig) is fed to
The ammonia-3 plant is designed for 1500 HP-Flash Vessel via hydraulic turbine, where it
MTPD (1653 STPD) based on natural gas steam is expanded to approx. 560 kPag (81 psig).
reforming technology. The plant is designed and During this process, mainly the dissolved inert
supplied by Uhde. The plant was commissioned gases (H2, N2, CH4 and CO) are liberated along-
in January 1997. with some CO2. These inert gases are then fed
into the fuel gas system of primary reformer or
CO2 product is removed from the process gas by can be burnt in flare.
BASF's OASE process (See Figure 1). The
solution used is an aqueous amine solution with The flashed OASE solution from HP flash
an activator, contains around 40% MDEA and vessel is expanded to approx. 25 kPag (3.6 psig)
balance water (by weight). The OASE process is into LP-flash vessel. In LP-Flash Vessel, the
characterized by both a physical and a chemical OASE solution is heated with countercurrent
scrubbing system. CO2 removal system consists overhead gas (CO2)/steam from the stripper to
of a two-stage absorber, a CO2 stripper, a HP release CO2 from the solution.
flash drum and a LP flash drum.
Semilean solution from bottom of the LP-Flash
Process gas from low temperature shift vessel splits into two streams. A part of semi-
conversion is treated in CO2 removal section. lean stream, i.e. approx. 14%, is fed by means of
Process gas enters at bottom of CO2 absorber stripper feed pumps via lean/semi-lean solution
and amine solution enters from top in two stages heat exchanger to the top of the stripper.
i.e. lean solution and semilean solution. To Balance (86%) of solution is fed to the middle
prevent loss of solution & activator by of the absorber by means of semi-lean solution
entrainment in process gas, top of absorber pump. This absorbs a major amount of CO2 in
column is provided with BFW washing trays. the absorber.

2011 235 AMMONIA TECHNICAL MANUAL

2012 24 AMMONIA TECHNICAL MANUAL

 In the stripper, the semi-lean solution is stripped damper sticky operation. The front end up to
with countercurrent steam generated in reboiler. High Temperature shift converter started on
In this manner, most of the CO2 is removed August 28, 2011.
from the solution. The regenerated lean OASE
solution leaves the stripper bottom, cooled in CO2 removal section was prepared for start-up
lean/rich solution heat exchanger followed by and solution circulation was established on
solution cooler and fed to the top of the absorber August 29, 2011. LP flash drum level was
by lean solution pump. showing 99% (normal 35-40%) and stripper
level was showing 99% (normal 75-80%). CO2
The CO2 & steam vapors from the top of the production was lined up at 11:50 hrs. After
stripper are fed into the bottom of the LP-Flash lining up of CO2 product, LP flash drum level
vessel where vapors are used for heating up the (95-99%), stripper DP, LP flash drum DP, CO2
rich OASE solution, resulting in a better header pressure was fluctuating. Semilean
performance of the low pressure flash. To solution pumps suction pressure was also
prevent loss of solution & activator by fluctuating between 280 kPag (40.6 psig) to 330
entrainment with saturated CO2 gas, top of LP kPag (47.8 psig). CO2 header pressure was set at
flash column is provided with washing trays 31.6 kPag (4.6 psig).
using condensate recovered from CO2 product
or DM water during start-up. At 14:25 hrs, set point of CO2 header pressure
reduced to 26.6 kPag (3.8 psig). At 14:38 hrs,
The saturated CO2 product from LP flash drum semilean solution flow dropped below trip
top after cooling in cooler and condensate setting (<900 m3/hr – 3963 USGPM) and
separation exported to urea plant. semilean pump tripped. Stripper feed pump flow
also reduced substantially to 75 m3/hr (330
LP flash drum is a packed bed column with bed USGPM). Both lean and stripper feed pumps
height of 6.0 meter (19.7 ft) and total packing stopped. Solution strength was measured and
volume of around 137 m3 (4873 ft3). Top 1 found drastically reduced to around 17.4% from
meter (3.3 ft) of packing is provided with 2” SS 38.7% of normal MDEA. Level in the LP flash
pall rings and bottom 5 meter (16.4 ft) of bed and stripper reduced by solution draining to slop
provided with 2” CS pall rings. The vessel is tank. OASE solution circulation again started
situated on top of the HP flash drum. LP flash with lean solution pump and stripper feed pump
drum sump solution outlet elevation is approx. at 19:29 hrs with less flow.
18 meter (59 ft) and packed bed support is
approx. 29 meters (95 ft) above the ground At 19:35 hrs, attempts were made to restart
level. Solution outlet 28” line (approx. 10 meter semilean pumps but each attempt resulted in
long - 32.8 ft) divides into two branches; one pump tripping due to low flow. Sharp reduction
10” supplying solution to stripper feed pump in suction pressure observed from 250 kPag
(approx. 35 meter long -114.8 ft) and another (36.2 psig) to 47 kPag (6.8 psig) just after the
24” supplying solution to semilean solution pump start. Solution circulation stopped again.
pump (approx. 21 meter long – 69 ft). It was decided to open the suction strainers of
pumps.
The incident
Preliminary Inspection:
On August 28, 2011 at 00:34 hrs, plant tripped
due to 11kv power supply failure. After Both semilean and stripper feed pumps suction
confirmation of power availability, plant start- strainers were opened. Strainers of semilean and
up was delayed due to reformer FD fan suction stripper feed pumps were found fully chocked

AMMONIA TECHNICAL MANUAL 236 2011

2012 25 AMMONIA TECHNICAL MANUAL

 with 2” pall ring packing. This has confirmed  To remove damaged packing from bottom
disturbance in the packed bed of LP flash drum sump 50 m3 (1766 ft3). To reload remaining
or stripper. Both stripper and LP flash vessels packing and new 9 m3 (318 ft3) available SS
opened to inspect the bed condition. Summary pall ring packing to LP flash drum top.
of observations follows:  To repair top distributor riser screens and
de-shaped plates of feed flash gallery.
 50 m3 (1766 ft3) of CS pall rings found in LP  To remove packing from LP flash drum
flash drum bottom sump (See Figure 2). outlet 28” line, semi-lean pump 24” suction
 LP flash drum two bottom support plates of line and stripper feed pump 10” suction line
packing support grid found opened / lifted by filling and draining of water or by water
from the shell support ring near stripper hydro-jet/fire water jet in pipes. If no
steam inlet nozzle (N8) (See Figure 3). success in above two methods then by
Packing at bed support opening plugged due cutting the pipelines at appropriate locations
to packed bed height. Some clearances for ease of removal of packing. Vacuuming
observed in the support grid segments. was not considered for packing removal
 Top liquid distributor gallery plates found based on previous unloading experience
de-shaped (bent) and cracked near the bolts without any success, long pipes and higher
towards same side of damaged bed support elevation of LP flash drum.
due to vertically upward force of vapors.
 Top liquid distributor riser screens found Jobs Performed:
damaged and packings found on top of the
distributor gallery. (See Figure 4)  After lining up of contract, bottom 50 m3
 Some pall rings also found in stripper vapor (1766 ft3) of packing unloaded manually in
inlet nozzle to LP flash drum bottom. 24 hours from LP flash drum sump.
 Semilean solution Pumps and stripper feed  Scaffolding erected at LP flash drum bottom
pumps suction lines from LP flash drum for closed visual inspection and to finalize
sump found chocked with packing. See the repaired procedure. It was observed that
Figure 5 for LP flash drum outlet suction 2nd and 3rd packed bed support plates got
piping layout. lifted by approx. 8” from the shell support
 Stripper top & bottom man-ways opened ring due to vertical upward force.
and no abnormalities found in vessel. Adjustment of lifted plates back into
 HP Flash Drum also inspected from both top position was not easy task (safety concern)
and bottom man ways. No abnormalities due to packing weight on the support bed. It
found in the vessel. was decided to repair lifted plates in position
by covering the bed openings. Accordingly,
Actions Planned: the opening in the two lifted plates was
closed by welding perforated sheet box in
An emergency meeting was called to decide sections using SS316L (See Figure 6). It was
further course of actions. It was decided planned to do permanent repairs during
 To unload LP flash drum sump area 50 m3 planned inspection shutdown after 6 months.
(1766 ft3) packing.  LP flash drum packed bed bottom support
 To erect scaffolding for close visual plates secured with additional reinforcement
inspection of LP flash drum packed bed by tacking support from main beam to shell
support. support ring.
 To repair bed support grid from bottom of  No entry permitted at LP flash drum top to
the bed. Repair procedure to be finalized avoid any vibration and further loss of
after close visual inspection. packing from the packed bed until

2011 237 AMMONIA TECHNICAL MANUAL

2012 26 AMMONIA TECHNICAL MANUAL

 completion of sump area packing unloading  Lean, semilean and stripper feed pumps of
& repair of packing bed support lifted plates. CO2 recovery section started and OASE
 Top liquid distributor riser screens and de- solution circulation established on August
shaped liquid gallery plates opened and 29. Column levels were around 70% in
repaired in workshop. absorber, 74% in HP flash vessel (normal
 Damaged packing removed from LP flash 55-60%), 95-99% in LP flash vessel (normal
vessel sump unloaded packing and rest 35-40%) and 100% in the stripper (normal
approx. 41 m3 (1448 ft3) of the packing 75-90%). Column levels were not adjusted
reloaded in the packed bed top. after establishing the OASE circulation.
 New 9 m3 (318 ft3) 2” SS pall ring packing  09:32 hrs to 11:00 hrs: solution heating was
loaded above 2” CS pall rings. Top SS bed completed, process gas introduced to
height was now 400 mm (1.3 ft) against absorber and CO2 product to header / urea
original design of 1000 mm (3.3 ft). plant lined up.
 Stripper feed pump suction isolation valves  Semilean solution pump suction pressure
spool dropped for ease of removal of was higher and between 310-320 kPag (45-
packing. 46.4 psig). During normal operation, semi-
 Semilean solution suction strainers kept lean pump suction pressure remains around
opened and secured with plastic and wooden 230 kPag (33.3 psig). Higher suction
sheet to prevent any debris towards the pressure indicates the solution level was up
pump. to 1 to 2 meter (3.3 to 6.5 ft) into the packed
 Packing removal tried from LP flash drum bed. This is around 1.6 to 2.6 meter (5.2 to
bottom nozzle by hydroject and fire water 8.5 ft) above the steam inlet nozzle from the
jet without any success. In order to save stripper.
time, LP flash drum bottom solution outlet  LP flash drum was provided with only one
line was cut at 3 locations. One at T-joint of level indicator (not controller). Level in LP
both pumps suction, one at 24” semilean flash vessel was fluctuating between 95-
pump suction line and one at 10” stripper 99%. Level was not showing higher than
feed pump suction line. 99% due to calibration error.
 Packing unloaded after cutting the pipeline  Physical location of LP flash drum level
manually and by using hydrojet water from transmitter LP taping is very near to steam
opened pipe access. inlet nozzle from stripper i.e. at a
 Boroscopic inspection done from circumferential distance of 2.5 meter (8.2 ft)
appropriate locations for both 24” and 10” and 0.25 meter (0.8 ft) below nozzle
pipeline to ensure that pipes were free of elevation. Lower indication (95-98%) of
packing. However, entire length of the 10” level was due to turbulence effect of steam
line could not be inspected using this entering in LP flash vessel high level.
method.  At 12:46 hrs semilean solution flow was
further increased from 2100 (9246 USGPM)
Problem Analysis: to 2400 m3/hr (10567 USGPM). LP flash
drum level came down to 87% due to
 After plant trip on August 28, 2011 process sudden withdrawal of liquid from LP flash
gas BFW wash water valve to CO2 absorber drum and pump suction pressure reduced to
and LP flash drum condensate return valve around 280 kPag (40.6 psig). Reduction in
were not isolated. DM water make-up pump suction pressure reconfirming the
continued in the LP flash drum during level in LP flash drum into the packed bed.
shutdown period of 24 hours. This increased  Semilean solution flow is a closed
the level in the columns during start-up. circulation of solution between absorber, HP

AMMONIA TECHNICAL MANUAL 238 2011

2012 27 AMMONIA TECHNICAL MANUAL

 flash and LP flash vessel. After increasing USGPM) to less than 900 m3/hr (3963
the semilean solution flow rate at 12:46 hrs USGPM).
to 2400 m3/hr, absorber and HP flash drum  Semi-lean pump tripped due to low flow
level was maintained at set-points by level caused by low suction pressure. Stripper
controllers. But LP Flash drum level again feed pump flow also came down. This
increased between 12:55 hrs to 14:24 hrs i.e. indicates restriction / blockage in the suction
Semilean solution pump suction pressure line of both the pumps.
increased from 280 kPag (40.6 psig) to 330  In summary, packing bed support failed due
kPag (46.4 psig). to lining up of stripper vapors in LP flash
 Due to increase in LP flash vessel level drum having a high liquid level i.e. level
above steam inlet nozzle, overhead steam higher than vapor inlet nozzle.
from stripper condensed in LP flash drum
bottom and CO2 product vent opening Post Repairs Experience:
reduced to zero between 13:30 hrs to 14:24
hrs. After completion of all repair jobs as listed in
 Pump suction pressure of 330 kPag (47.8 jobs performed, LP flash drum was boxed up on
psig) indicated that packing bed filled with September 4, 2011 at 18:30 hrs. On September
solution up to 3.0 meter (9.8 ft) (approx. 3.6 5, 2011 solution circulation was established and
meter – 11.8 ft above the steam inlet nozzle CO2 product lined up to Urea plant. Ammonia
from the stripper). production was lined up on September 6, 2011
 At 14:25 hrs, CO2 header pressure came at 10:24 hrs. It was noticed that stripper feed
down to 30 kPag (4.4 psig). Set point of CO2 pump suction pressure came down to 50 kPag
header pressure reduced from 31.6 kPag (4.6 (7.3 psig) (instead of normal 250 kPag – 36.2
psig) to 26.6 kPag (3.9 psig). CO2 vent psig) and flow restricted to 250 m3/hr (1100
opened immediately and vapors started USGPM) against normal flow of 380 m3/hr
escaping through minimum resistance path (1673 USGPM). Suction strainer of both the
in LP flash drum very near to steam inlet pumps were opened one by one and packing
nozzle. The vapor slugged (developed again found in strainers. Even after removal of
vertically upward force) through the liquid packing from strainer suction pressure improved
lifted two trays off from their support. The to 100 kPag (14.5 psig). However, suction
slugging of vapors through the liquid had pressure of semilean solution pump remained
also bent upward liquid gallery trays and normal at around 230 kPag (33.3 psig). This was
displaced liquid distributor riser screens. confirmed that 10” stripper feed pump suction
Due to damage of vapor riser screens, pieces line was still left with some packing. Location
of packing from bed were blown out and of chocked packing was confirmed in pipe
found on top of liquid distributor and through X-ray. It was identified that packing
gallery. from stripper feed pump common suction line
 Due to high level in LP flash drum and high near the pumps (approx. 10 meter -32.8 ft) were
downward liquid flow, packing started not removed (See Figure 6).
coming out from bed opening and escaped
in semilean and stripper feed pumps suction The following was done to remove the packing
lines. Immediately pump suction pressure from line.
started reducing from 330 kPag (47.9 psig)
to less than 50 kPag (7.3 psig) at 1435 hrs.  Frequent changeover of the stripper feed
Pump discharge flow also reduced pump was done to increase sudden flow
accordingly from 2434 m3/hr (10716 through pipe, but after some attempts,

2011 239 AMMONIA TECHNICAL MANUAL

2012 28 AMMONIA TECHNICAL MANUAL

 packing stopped coming out in strainers as  A packing holder/cage implementation is
badly chocked in pipe & elbows. also planned during annual shutdown of year
 Packing removal was tried with light 2015 around the bottom sump vortex
wooden hammering on the chocked pipe breaker of LP flash drum to arrest packing in
locations to release the chocking with flow case of bed support failure.
of solution. This started release of chocked
packing but became stuck in the pump Conclusion:
suction isolation valve seat and valve started
passing badly. Due to this pump changeover  Condensate / DM water recycle to LP flash
for suction strainer cleaning was not drum and CO2 absorber top BFW wash
possible. water was not isolated after trip. This diluted
the solution and increased level in the CO2
Plant production was reduced by 14% compared removal system columns before the start-up.
to load prior to packed bed failure incident due  CO2 recovery system: solution heating by
to reduction in lean solution flow. To avoid reboiler was lined up before liquid level
production loss, hot bypass was made on suction adjustment in LP flash drum. Heating started
line (chocked location already identified with high level in LP flash drum; level was
through X-ray). Hot taping was made just after above the steam inlet nozzle from stripper.
T-joint in 10” suction line and connected This increased the backpressure in stripper
downstream to standby stripper feed pump and increased reboiler temperature.
suction isolation valve (See Figure 5 marked  High semilean pump suction pressure and
with hot bypass solution line). After lining up of reboiler saturation temperature confirmed
bypass line, suction pressure improved to 250 the high level in LP flash drum.
kPag (36.2 psig) and normal flow established  Lining up of stripper overhead vapors along-
through the pump. Ammonia production with high level (liquid level in packed bed)
improved by 13.3% due to increased lean in the LP flash drum slugged stripper vapor
solution flow. through liquid near the steam inlet line.
Vapor force (driving force) again increased
Semilean solution CO2 loading was observed due to depressurization of the column
increased due to minor channeling after the caused by reduction in CO2 header pressure.
incident of LP flash drum bed failure. CO2 slip This developed a high vertical force and
from absorber increased accordingly and lifted packing bed support plates from shell
ammonia production reduced by approx. 0.7%. support ring.
 Due to high vertical force, upper distributor
Long term repairs: gallery plate was damaged (found bent), gas
risers screen also damaged on the same side
 As inspection shutdown was already of support bed lifting. Damaged screens and
planned in March, 2012, it was planned to blown out packing from bed were found on
replace complete LP flash drum packing top of liquid distributor gallery.
with 2” IMTP SS304 from 2” CS pall rings.  Lack of packing holders at LP flash drum
Bed support replacement also planned bottom caused all pumps suction pipelines to
during the shutdown window. Both LP flash be chocked with packing rings.
drum packing and bed support replaced
during inspection shutdown of March, 2012 Recommendations:
as per plan. Packing was removed from 10”
pipe after cutting from two locations during  Packings are vulnerable to damage during
the shutdown. start-up and shutdown operation. Base liquid

AMMONIA TECHNICAL MANUAL 240 2011

2012 29 AMMONIA TECHNICAL MANUAL

 level rise above reboiler return or vapor feed  In case of abnormal increase in DP of the
nozzle has been one of the most common columns, cut-off vapors / gas streams
cause of packing failure. Bottom liquid level immediately to avoid internal damage.
needs to be watched closely during the start-  Avoid pressure fluctuations in packed bed
up. during start-up and particularly when liquid
 Keep liquid level in columns in operational level is high. Depressurization of the column
range during start-up / normal operation. If to be done with precaution to avoid damage
high, reduce level in columns by draining (in of bed support by sudden release of
slop tank) during initial circulation and excessive vapor flow through the bed.
before lining up of steam / gas stream in  It is recommended to provide packing holder
packed bed columns to avoid vapor in LP flash drum bottom surrounding vortex
slugging through liquid. breaker to avoid cutting and welding of
 If column operates with only one level pipes in case of similar incident.
indicator, do not rely on transmitter when
level is above 95%. Observe other Acknowledgement:
parameters like pressure drop, column
temperatures, pumps suction pressure etc. to Our sincere thanks go to Mr. Ali Jama, Qafco 5
understand abnormal operation. & 6 Commissioning Manager, Mr. Richard de
 If applicable use pumps suction pressure as La Bastide, Production Manager (Q1 to Q4) and
reference and provide pump high suction Mr. Iftikhar Turi, Overall Oncall Coordinator
pressure alarm to avoid even faulty level for their guidance and support.
indication.

Figure 1: CO2 removal process

Figure 1: CO2 removal process

2011 241 AMMONIA TECHNICAL MANUAL

2012 30 AMMONIA TECHNICAL MANUAL

 Figure 2: Packing in LP flash drum Sump Area

Figure 3: LP flash drum damaged bed support plate

AMMONIA TECHNICAL MANUAL 242 2011

2012 31 AMMONIA TECHNICAL MANUAL

 Figure 4: Top side liquid gallery damage

Figure 5: LP flash drum outlet piping

2011 243 AMMONIA TECHNICAL MANUAL

2012 32 AMMONIA TECHNICAL MANUAL

 Figure 6: Packed bed Support Plate Repair

AMMONIA TECHNICAL MANUAL 244 2011

2012 33 AMMONIA TECHNICAL MANUAL