Failure of Inner Shell of Double

Walled Atmospheric Ammonia Tank

A 5,000 MT atmospheric ammonia storage tank, which was originally commissioned in March
1998, needed to be decommissioned and repaired due to the failure of the inner shell of this double
watted tank. The causes of the failure, as well as the corrective and preventive actions taken, are
discussed.

K. A. Nair and N. V. Jagan Mohan

Coromandel Fertilizers Ltd., Visakhapatnam, India

Introduction 1 metric ton =1.1 short tons

l m = 3.28 ft

C
oromandel Fertilizers Limited (CFL) was I km = 3,280 ft or 0.62 miles
established as an integrated fertilizer plant, in 1°C = 1.8°F
1967, to manufacture various grades of 1 m3 = 1.3079 cubic yards.
NP/NPK fertilizer. The site initially had a 360 TPD
ammonia plant, 400 TPD urea plant, 600 TPD sulfuric Background of Incident
acid plant, 255 TPD phosphoric acid plant and two
identical 550 TPD granulation trains. The plant is At the inception, manufactured ammonia was stored
located on a 480 acre site and has a captive berth at the in 2 x 1,500 MT Horton spheres, operating at 2.3
port for handling cargo. After 33 years of operation, kg/C2g. In view of the expansion of production capac-
the naphtha based ammonia/urea plants have been ity, and the decision to gradually phase out the aging
closed down due to high feedstock prices, and ammo- Horton spheres, it was decided in 1996 to put up a
nia requirements are now met through a 5.5 KM import 5,000 MT double wall, double integrity atmospheric
pipeline between our jetty and ammonia storage facili- ammonia tank, along with refrigeration compressors.
ty. Urea is also outsourced, while sulfuric acid produc- This facility was completed in March 1998.
tion is currently 1,200 TPD, and phosphoric acid 450 The shell plates, bottom plates, anchor strips, and
TPD, to supply most of the acid requirements for mak- fasteners are all of low temperature carbon steel
ing 600,000 TPY of 28:28:0, 14:35:14 and 20:20:0 (LTCS-ASTM 517 or equivalent) whereas SS304 was
grade fertilizers. used for the expansion bellows. All piping used was of
Between April 2000 to March 2001, our company seamless LTCS. Further details are given in Table 1.
had a sales turnover of approximately $135 million, During fabrication, all joints were tested by
with a profit of $15.6 million before tax. Radiography/DP. After completion, the tank and asso-
Note: We have used metric units. Typical conver- ciated fittings were inspected per API-620, with an air
sions are: test up to 1,750 MM WC for the tank and cup, after
1 kg/cm2 = 0.967 atmospheres (0.980 bar) water filling for the inner cup hydro test, followed by

AMMONIA TECHNICAL MANUAL 47 2002

 Table 1. Atmospheric Ammonia Storage Tank

DESIGN CODE
SHELL & BOTTOM : a) API 620 APPENDIX 'R' (JUNE 1990)
b) API 650 (JULY 1993) FOR SHELL STABILITY

ROOF STRUCTURE : DIN 41 19 SHEET 1 & 2 (OCTOBER 1961)

COMPRESSION RING & ROOF

PLATE : API 620 (JUNE 1990)

WIND PRESSURE IS 875 - 1987

SEISMIC DESIGN : IS 1893 - 1991

MANUFACTURING CODE : API 620 (JUNE 90)

DESIGN DATA

DESIGN PRESSURE mmWC : + 1400 AND -75

DESIGN TEMPERATURE 'C :-35

OPERATING PRESSURE mmWC : 700 -1000

OPERATING TEMP .(MAXIMUM)C : -32.5

CORROSION ALLOWANCE mm : NIL : 3 mm. TOTAL CORROSION ALLOW.

FOR ANCHORS EMBEDDED IN
CONCRETE ONLY

PROCESS FLUID : AMMONIA (LIQUID & GAS)

DENSITY OF FLUID STORED Kg/ M3 : 680.8 at - 32.5 C and 683.9 at - 35 C

MAX.LIQ .LEVEL IN OPERATING

CONDITION M (IN CUP SHELL ) :19.8

EFFECTIVE VOLUME M3 : 7364.85 EFFECTIVE VOLUME BETWEEN 50mm

ABOVE TOP OF NOZZLE N2 AND
MAXIMUM OPERATING LIQUID LEVEL

AMMONIA TECHNICAL MANUAL 48 2002

 Table 2. Atmospheric Ammonia Storage Tank

AMMONIAC/VRGODE1FAILS CFL RECEIPT VS.SHIP DISCHARGE

SHIP DATEOFSHIP SHIP NAME INITIAL FINAL OPENING STOCK SHIP DISCHARGE QTYRECD

NO. UNLOADING LEVEL LEVEL QUANTITY QUANTITY BY CFL

nm mm MT MT MT
1 18.4.99 M.T.HECTOR 607 15060 157 3771 3774

2 15.5.99 BRUGGE VENTURE 3355 19433 869 4196 4197

3 03.6.99 BRUGGE VENTURE 2357 19152 610 4198 4197

4 27.6.99 M.V.DONAU 3391 17980 878 3500 3479

5 11.7.99 HERAKLES OSLO 4115 19070 1066 3165 3150

6 08.8.99 MV.HAVJARL 2140 16400 554 3910 3913

7 18.8.99 GAZKANDLA 2001 18800 518 4417 4416

8 11.9.99 GAZKANDLA 3791 9378 982 1528 1528

9 21.9.99 GAZKANDLA 813 16700 211 5050 4470

Photo 1. Prominent bulge in floor plate.

AMMONIA TECHNICAL MANUAL 49 2002

 Photo 2. Bulged portion leading up to inner wall.

Photo 3. Rupture near level indicator float chamber.

AMMONIA TECHNICAL MANUAL 50 2002

 Photo 4. Eight inch long tear.

Photo 5. Detailed view of the rupture.

AMMONIA TECHNICAL MANUAL 51 2002

vacuum test. All instruments were tested and calibrat- series indicator located on the control room panel
ed also. which also showed cup levels in percentage. The
In April 1998, work started on an ammonia import designer had provided a drain nozzle with valve, to be
facility consisting of unloading arm at the berth, and a kept normally blanked, but to be connected by a tem-
5.5 km cross-country pipeline of seamless LTCS up to porary line to the pump suction line, as and when the
the atmospheric storage tank. This was completed in annulus was to be drained. While normally there would
March 1999, and the first ammonia ship was unloaded be little chance of condensation in the annulus due to
in April 1999. The very next month, the ammonia plant high ambient temperature, draining could be necessi-
was closed down. tated following an overflow of the cup.
Between April and September, a total of 9 consign- An investigation showed that the "as built" wiring
ments were received, varying in size between 1,500 to diagram for the above instrument was incorrect.
5,000 MT. While receiving large consignments into However, the instrument had been wired correctly by
the 5,000 MT tank, we were simultaneously pumping the contractor at the time of commissioning, and had
ammonia through a heater into the Horton spheres, to been operational until May 1999, by which tune 2 ships
ensure safe level in the tank at all times. Maximum ship had been unloaded. However, the contractor did not
unloading rate was 500 te/h (Table 2). correct the "as built" drawing handed over to us, along
After the 9th ship was unloaded on September 21st, with other commissioning documents. In May, after
1999, we observed a short receipt of 580 MT of ammo- some work on the instrument, it was reconnected by the
nia. Investigation showed an equivalent quantity to be technician as per the "as built" drawing. Since this was
present in the annular space, between the inner and incorrect, and the transmitter used was not of a "smart"
outer tank walls. It was quickly established that the type (such as Bells 761), the instrument read zero,
presence of ammonia in the outer space was due to rather than indicating a fault or negative reading.
communication between this and the inner cup (and not Based on the above indication, and in the absence of
due to an overflow from the inner cup during ship any specified procedure to routinely drain the annulus,
unloading) (Figure 1). no such action was initiated.
Subsequently, it appears that some ammonia had
investigation started collecting in the annulus, due to splashing from
the inner cup during unloading at high levels. From
Decommissioning of the tank was started from ship No. 7 onwards (in August), there also appeared to
October 14, 1999 and vessel entry was effected on have been some condensation in the annulus, due to the
November 3. An inspection showed that the lap weld- low temperature prevailing there. (Normally, the annu-
ed sketch plates at the bottom of the cup had bulged lus temperature, when dry, is close to ambient.) It
inward, by nearly 400 mm, over a width of 200 mm, appears that by September, prior to the arrival of the
and a length of several meters. Where the deformation ninth vessel, over 250 tes ammonia had accumulated hi
reached the circumferential inner shell, there was a 200 the annulus, corresponding to over 6.50 m of level.
mm bulge with a star shaped tear of 150 mm by 125 Since the ninth ship had a parcel of 5,050 MT of
mm (Photos 1 to 5). ammonia, the tank level was brought down to 813 mm
Removal of damaged plates along with the inspec- (around 211 MT) prior to receipt, this being the lowest
tion of the entire tank, covering all weld joints as well level ever reached after the first vessel was unloaded hi
as sketch and circumferential plates near the failure April. We also planned to pump out ammonia from the
zone, started November 4, onwards. Simultaneously, tank simultaneously while unloading, to ensure final
an investigation was made to identify the cause of the free board of 325 mm was maintained.
failure. When the cup level was reduced to .813 mm, failure
The level hi the annular space of the tank was sup- appears to have occurred due to the hydrostatic head of
posed to be measured by a level transmitter, which was ammonia in the annulus acting below the bottom plate,
provided with a high level alarm, with 100% level cor- leading to extensive plastic deformation of the plate,
responding to l m level, as shown on a Dual Bells 4300 ultimately resulting in the fracture of the bottom and

AMMONIA TECHNICAL MANUAL 52 2002

 Table 3. Atmospheric Ammonia Storage Tank Sequence of Events

"" ' : •*
Ï ;

; iStarting Completion
1 i Water fling j 14.10.99 19.10.99
2 ^Drahing of ammonia Iquorto naphtha tank [18.10.99 19.10.99
3 ;Open'ngof roof top manwaycover Î16.10.99 16.10.99
4 ; Nirogen purging i 21 .10.99 22.10.99
5 iAdrritting compressed air ;23,10.99 24.10.99
6 | Ope nhg of bottonVoutsr man hole cover j25.10.99 25.10.99
7 :Blcwerœnnectbn& disconnection ! 26.10.99 27.10.99
8 ; Inside ma nway removal ; 27 .10. 99 27.10.99
9 ; Blower connecfon&disconnectbn i27 .10.99 30.10.99
1 0 Wash hg and drain'n g of water 1 01 .1 1 . 9902.11.99
1 1 : Clean hg and drainh g th e wate r torn floor ! 03 .1 1 . 9903.11.99
1 2 : Cursory inspection i 30 .1 0.99 31.10.99
1 3 iPlatecuffing for initial inspection (Bottomplate) i04.11.99 08.11.99
14:Totalplateremoval i 09 .11. 99 16.11.99
15jShellrepair J24.11.99 04.12.99
16 Concrete Kerb repa r il9.11.99 21.11.99
17 ;Bottomplate inspection i20.11.99 20.11.99
1 8 ;Sand removal i15.11.99 20.11.99
19;Sandfling 1 21 .11. 99 25.11.99
20 Inspection of hside courses i05.11.99 03.12.99
21 ; Inspection of outside courses i 24 .11. 99 09.12.99
22';Bottomplatefixing ;27 .11.99 02.12.99
23 ; Bottom p late welding hspection -30.11. 99 03.12.99
24 : Vacuum box testing !01 .12.99 05.12.99
25; course plate ;04.12.99 05.12.99
26 ; Final hspection before water fling ; 06 .1 2.99 06.12.99
27 ilnside cup manway welding ;07 .12.99 07.12.99
28 Water fling ofcupshell J08.12.99 11.12.99
29 ;Tank shellmanway doshg i 12.12.99 12.12.99
30 .Roof manway & roof plate doshg i08.12.99 12.12.99
31 iPneumate testing of lank she! ';13.12.99 14.12.99
32 ; Reiefvalves testing & vacuum breaker testhg • 14.12.99 14.12.99
33 Water drainhg i 14.12.99 17.12.99
34 . Nirogen purging i 19.12.99 20.12.99
35 Nirogen bbnkethg i20.12.99 29.12.99
36 Ammonia vapours into the tank ;29.12.99
37 Ammonia liquid hto the tank '•_

AMMONIA TECHNICAL MANUAL 53 2002

 Figure 1. Atmospheric ammonia storage tank.

Figure 2. Atmospheric ammonia storage tank.

AMMONIA TECHNICAL MANUAL 54 2002

side plates as well as the circumferential weld (the fig- the drain pot every week was initiated.
ure of 250 MT accumulation is based on an estimate of (7) The siphon breaking arrangement on the liquid
the head required to buckle the floor plate). inlet line into the tank was modified, so that the top
most hole of the dip pipe is above the maximum liquid
Repair and Reeommissioning level in the tank (Figure 2)
(8) All documents have been updated to reflect "as
Repair work started on November 14, and was com- built" conditions and all other documents have been
pleted by November 30. It involved repair of the con- removed.
crete kerb, removal and refilling of bottom sand layer, (9) A routine for checking all instruments has been
repair and replacement of shell courses, bottom plate drawn up.
renewal, and other modifications. Afterwards, a thor-
ough inspection of the repaired areas as well as resid- Conclusion
ual zones was completed by December 6, including
vacuum box testing, radiography and ultrasonic testing. The tank has been in operation since January, 2000
Water filling of the inner cup was followed by nitro- without any other problems. However, to improve safe-
gen purging, and ammonia was introduced on ty and operating practices at site, we have also gone
December 29. The period between ammonia out and in ahead with the following:
was approximately 2.5 months (Table 3). (1) A new 7,500 MT capacity tank, which was
already contracted in April 1999, was commissioned in
Corrective and Preventive Action July, 2000. This will give us adequate storage capaci-
ty to handle 7,500 MT of ammonia imports a month,
(1) The tank was derated to 4,865 te capacity, setting without using the Horton spheres.
alarms and trips accordingly, so as to provide a mini- (2) An ISO document control system has been intro-
mum 500 mm free board between maximum allowable duced to ensure that only authenticated documents are
liquid level and overflow level (BS 7777, 1993). available to operators.
(Earlier, a free board of 325 mm was provided.) (3) Process Safety Management Systems has been
(2) A separate recording instrument was provided to implemented to ensure procedures such as manage-
record tank inner cup and annulus levels. A parallel ment of change, and pre-startup review, are institution-
independent level indication was also provided for the alized.
annulus. While contractors use the API-620 (APP-R) design
(3) All indicators were changed to fail safe (Bells code for designing ammonia storage tanks, we have
761) type instruments. found that BS 7777 better addresses some of the oper-
(4) A temperature indicator was provided to monitor ators concerns, such as the danger of liquid buildup in
annulus temperature, with a low temperature alarm. the annular space, the need for adequate free board to
(5) An interlock was provided to trip the ammonia avoid splashing and spill over, and so on. A separate
pumps if annulus level reaches 400 mm (a safe level international code of practice covering aspects of
considering the "dead" volume on the inner cup). design, construction, installation and operation of such
(6) The Annulus drain was permanently connected to tanks would ensure the contractor and the operator bet-
the drain pot, and a routine of draining the annulus into ter understand each others requirements.

AMMONIA TECHNICAL MANUAL 55 2002