Successful inspection of two large Ammonia Storage Tanks · Ltd. The tanks were commissioned by Snamprogetti in 1985. The storage tanks are cylindrical with domed roofs and designed

Successful inspection of two largeAmmonia Storage Tanks

GPIC complex has two atmospheric ammonia storage tanks, each of 20,000 MT capacity. The tankshave been in operation since 1985. European Fertilizers Manufacturing Association guidelines for

‘Ammonia Tanks Inspection’ were adopted to set the frequency for inspection of the tanks.

One ammonia tank was decommissioned in 2004 for internal inspection as well as replacement ofinsulation blocks. The decommissioning, inspection and recommissioning project duration was 5

months. Based on this experience, the 2nd tank was successfully decommissioned and inspected in theyear 2006 in just 4 months.

This paper describes the methodology followed for the inspection project of the tanks along withdetails of decommissioning and recommissioning.

Yusuf AbdullaGulf Petrochemical Industries Co.

Kingdom of Bahrain

Yoga NarasimhaGulf Petrochemical Industries Co.

Kingdom of Bahrain

1.0 Introduction

ulf Petrochemical Industries Co. (GPIC)was formed on 5th December 1979 as anequal partnership between the

governments of Bahrain, Saudi Arabia andKuwait with the primary objective of utilisingBahrain’s natural gas for the production ofpetrochemicals.

GPIC’s first project was its grassrootpetrochemical complex at Sitra Island on thenorth-east coast of Bahrain. A site of 60 hectarewas successfully reclaimed from the sea toconstruct the ammonia and methanol plants(with an original capacity of 1000 MT per day

of each product), along with the related Utilitiesand Offsite facilities. In 1989, the ammonia andmethanol plants were debottlenecked to aproduction level of 1,200 MT per day each. In1998, GPIC diversified into the manufacture offertilizer with the commissioning of a 1,700 MTper day single stream Granular Urea plant withassociated Offsite facilities.

2.0 Description of the NH3 tanks

The GPIC complex has two ammonia storagetanks (T-7101A and B) each of 20,000 MTcapacity for storage of liquid ammonia at nearatmospheric pressure. Basic engineering for theammonia tanks was done by UHDE GmbH and

G

2412007 AMMONIA TECHNICAL MANUAL

construction by Motherwell Bridge EngineeringLtd. The tanks were commissioned bySnamprogetti in 1985.

The storage tanks are cylindrical with domedroofs and designed as double wall doubleintegrity tanks (cup in tank). Each tank isinstalled on a concrete raft foundation protected

by an electrical heating system which preventsfreezing of the load bearing insulation blocksand the subsoil below (Refer to Fig.1). Thetanks are externally insulated with polyurethanefoam (PUF). The aluminium cladding joints onthe roof and sidewalls of the tanks are vapoursealed. The technical data of the ammonia tanksare given in Table-1.

BASE INSULATION DETAIL FOR AMMONIA TA NK

A S PER SNAMPROG ETTI'S SPECIFICATIONS

NOTES:-

DETAIL-A

BASE INSULATION DETAIL FOR AMMONIA TANK

DETAIL A

NOTES

AS PER SNAMPROGETTI’S SPECS.

BASE INSULATION DETAIL FOR AMMONIA TA NK

A S PER SNAMPROG ETTI'S SPECIFICATIONS

NOTES:-

DETAIL-A

BASE INSULATION DETAIL FOR AMMONIA TANK

DETAIL A

NOTES

AS PER SNAMPROGETTI’S SPECS.

Figure-1 : Tank Foundation Base Insulation Arrangement

The ammonia storage facility has three screwtype compressors for maintaining the tankpressure. Three high capacity ammonia pumpsare available for export to ships. A separatevapour return line has been provided to handlevapours from the liquid loading line as well asfrom ships. A common flare system is providedfor both the tanks.

3.0 Operation and MaintenanceHistory

The ammonia storage tanks have been inoperation since the year 1985. The performance

of the storage tanks and the refrigeration systemhas been satisfactory due to the stable operationof the tanks between 40 and 50 mbarg pressure.The base heaters of the tanks have beenoperational throughout and an effectivepreventive maintenance schedule has been inforce.In 1995, both the tanks showed signs ofdeterioration of the roof and shell vapour seals.The insulation was found saturated with water atsome locations on T-7101B. Complete renewalof the vapour sealing was carried out for boththe tanks T-7101A and B. Again in 2002, thevapour sealing system for both the tanks showeddeterioration and was renewed.

242 2007AMMONIA TECHNICAL MANUAL

Type Double wall double integrity cup in tank with dome roof

Design code API 620 Appendix ‘R’

Storage capacity 20,000 MT

Design temperature -33oC

Design pressure 100 mbarg

Tank diameter 41.6 m (outer shell); 40.0 m (inner shell)

Cylindrical height 25.4 m (outer shell); 24.8 m (inner shell)

Wall thickness (inner/outer shell) 23 mm (bottom tier); 8 mm (top tier)

Roof thickness 5 mm

Foundation Concrete raft foundation.

Shell / Roof insulation 100 mm PU-Foam with 0.8 mm aluminium cladding

Bottom insulation 50 mm sand screed layer + 2 layers of foamed glass 100 mm and 200 mm insulation blocks

Base heating elements 46 elements divided into two zones (2 thermocouples in each zone for measurement)

Base heater temperature indication One for each zone

Table-1 : Technical Data of ammonia Tanks

The stairway platform supports on the top forboth the tanks were found to be corroded andthese were renewed. To ascertain the conditionof the tank foundations, the concrete basedinsulation blocks below tank T-7101B wereinspected in September 2002. This inspectionrevealed severe icing and varying degrees ofdeterioration on the outer concrete insulationblocks (refer to Fig. 2).

Figure-2 : Icing on Outer Insulation Blocks

A similar icing phenomenon was observed intank T-7101A. To determine the failuremechanism of the concrete insulation blocks inthe foundations of tanks T-7101A and B,specialists from UHDE and Motherwell Bridgewere invited to site for inspection.The icing and deterioration phenomenon werereviewed. Core samples were collected from adepth of approximately 400 mm from theinsulation blocks of T-7101B for analysis andcompared to design specifications. Thelaboratory results of the core sample are shownin Table-2.

Analysis Actual Allowable Limit

Moisture Content (%) 85 – 90 < 12

Compressive strength N/mm2 1.5 – 2 4 – 6

Density Kg/m3 500 – 550 400 – 500

Chlorides % 0.03 0.06

Table-2 : Core Sample Analysis


The analysis revealed significant damage to theouter ring of insulation blocks. This wasattributed to moisture ingress through‘breathing’ of air into the blocks. Thecompressive strength was also lower thanspecification. This ‘breathing’ effect could bedue to the damage to the vapour seal on the tankshell despite complete replacement in 1995. Allthe core samples showed severe icing at theouter edge of each sample.As a temporary measure to prevent furtheringress of moisture, a skirt made of aluminiumcladding sheets was fixed to the bottom of theshell insulation covering the concretefoundation to prevent moisture ingress into thefoundation and to divert condensation and rainaway from the foundation blocks.

4.0 The Reasons for the TankInspection

Based on analysis of the insulation block coresamples and site inspection findings andrecommendations, replacement of the damagedbottom insulation blocks was planned.

Applying European Fertilizer Manufacturer’sAssociation (EFMA) “Guidelines on AmmoniaTanks Inspections” to GPIC’s ammonia tanks,the inspection frequency falls between 10 and20 years. The tanks had not been inspectedsince commissioning.

GPIC management approved plans todecommission and inspect one tank in 2004. Itwas deemed necessary to carry out the internalinspection of the tank T-7101B along withreplacement of the insulation blocks. Thefollowing broad strategy was adopted:

Ø Review and finalise proposals for thereplacement of insulation blocks.

Ø Firm up inspection scope based onEFMA guidelines.

Ø Finalise decommissioning and recommissioning procedures for review

by UHDE.

Ø Start decommissioning of T-7101B inApril 2004 and replace insulation blocksunder UHDE supervision, inspect andrecommission the tank.

Ø Ensure stability and monitorperformance of T-7101B for 2 years;then proceed with inspection of T-7101A in 2006.

On the advice of UHDE, it was decided toreplace the damaged insulation blocks byinstalling wooden wedges made up of aspecified type of hard wood.

5.0 Decommissioning Approachand Methodology

5.1 Options for decommissioning:

The complete list of activities was sequenced inline with the decommissioning andrecommissioning operations. Followingmaximum reduction of the ammonia level in thetank, three options were considered for removalof the liquefied ammonia below the pumpsuction.

1) Natural evaporation.2) Allowing the liquid to evaporate by

admitting only hot ammonia vapour.3) Adding water to the tank and pumping

out the aqueous ammonia.

Natural evaporation was discarded as too timeconsuming. Water addition was also discardeddue to associated waste disposal problem. Theoption of hot ammonia vapour addition wasselected as it is environmentally friendly and theoverall duration was within the required timeschedule.

5.2 Project Team Formation:

A dedicated project team was formed in year2003. The procedures for decommissioning andrecommissioning were prepared and reviewedby UHDE. P&I diagrams for bothdecommissioning and recommissioning


situations were developed and subjected to adetailed Hazop study.

5.3 Need for supporting the tank:One critical aspect was the need for supportingthe tank prior to warming up of the interior asthe damaged bottom insulation blocks wouldlose their strength while thawing. Based onstatic load calculations, it was decided toremove the insulation blocks between theanchor point locations to support the tank priorto thawing. The tank has been provided with 65anchor points on the outer shell. These areembedded in the concrete raft foundation. Onlyafter supporting the tank between anchor points,warming of the interior could be started, alongwith the removal of insulation blocks in theremaining locations and replacing these withwooden wedges.

5.4 Risk Analysis / Environment Impact:

The hazards associated with the project wereidentified by the team and an appropriate riskanalysis plan was prepared. Suitable mitigationmeasures were also included to reduce theoverall risk to an acceptable level.

The team also studied the environmental impactof the project. One important point thatemerged from this study was the impact ofmercury in the liquid ammonia. The presenceof mercury had first been observed at GPICduring 1996 at the drain point of the heatexchanger in the ammonia refrigeration loop.The natural gas feed stock to the GPIC complexnormally contains 50 microgram/Nm3 ofmercury. A guard reactor was installed in 1996to remove mercury in the feed natural gas. Thisprompted the team to consider the possiblepresence of mercury which could have collectedthere during plant operations before theinstallation of the mercury guard reactor.

Based on the experience of other plants andfollowing UHDE’s guidelines, suitableprocedures were developed for dealing with

mercury, both to eliminate potential explosivityas well as dealing with disposal. To enableanalysis for mercury in the vapor phase duringpurging operations, a portable analyser(Mercury Tracer 3000) was procured.

6.0 Temporary Modifications forDecommissioning

6.1 Emptying the tank:

In accordance with the decommissioningprocedure, the level in the tank would bereduced to around 1000 mm using the normalexport pumps, leaving approximately 900 MTof liquefied ammonia. To reduce the levelfurther, a temporary pump would be connectedto the 6” (150 mm) drain line of the tank. Thiswould enable reduction of the level to 450 mmcorresponding to the drain line elevation (Referto Schematic-1). For the final stage ofemptying, a knock-out drum would be providedbetween the tank and the pump. The knock-outdrum would be connected to the refrigerationcompressor circuit. By creating a suitablepressure differential between the knock-outdrum and the tank, siphoning action wouldenable liquid to flow from the tank, thusfacilitating the removal of the remainingquantity of liquid ammonia. The final liquidlevel would be approximately 50 mm whichrepresents the protrusion of the drain line abovethe tank bottom plate.

Schematic -1: Tank Drain Nozzle Elevation

1524 mm

300 mm50 mm

450 mm

T-7101A/B

Pump suction (24'')

Tank drain (6'')


The knock out drum, transfer pump along withrelated piping and instrumentation arrangements

are shown in Figure-3.

Figure-3 : Knock-out Drum and Pump Arrangement

6.2 Provision of 6” (150mm) vent inter-connection to flare:

As part of decommissioning sequence, the tankhad to be purged with nitrogen to expel allammonia vapour. During the initial phase ofpurging, the ammonia vapours would berecovered by the refrigeration compressors. Asthe ammonia content in the vapour decreased,the vapour would be diverted to flare. Toachieve this, the ammonia vapour from tankT-7101A would be isolated, as both tanks areinterconnected, followed by stopping therefrigeration compressors. Moreover theestimated duration for purging would be 10 daysand in that period the refrigeration system had tobe available for pressure control of tankT-7101A. To accommodate both purging ofT-7101B and pressure control of T-7101A, thevapour return line from the export wharf wouldbe disconnected from T-7101B and a suitable 6”(150mm) connection would be made to the flaresystem. This connection would facilitatetrouble-free purging of T-7101B without anyinterruption. The refrigeration system wouldremain available for T-7101A throughout.

7.0 Decommissioning of TankT-7101B

Based on the availability of the wooden wedgesand ammonia export shipping program, theinspection was scheduled for the first quarter of2004. All the required temporary pipingsystems with the above modifications fordecommissioning were fabricated and installed.The decommissioning activities started on10/04/2004.

7.1 Level reduction:

The main loading pump was started transferringfrom T-7101B to T-7101A and the level inT-7101B level was reduced to 1000 mm whichwas considered as a safe level for this highcapacity pump. Further level reduction to600mm was achieved using the temporary vanepump, which was connected to the 6” (150mm)drain of T-7101B. The vane pump dischargedto the 6” (150mm) drain line of the tankT-7101A. The pump capacity was ~30 MT/hr.The knock-out drum which was installedbetween the tank and the temporary pump was


taken in service. The temporary pump wasconnected to the knock-out drum for finalemptying.

7.2 Wooden wedging: (under coldconditions)

Two specialists from UHDE were on site tosupervise the wedging. As planned, removal ofthe insulation blocks between the anchor pointsstarted, followed by insertion of the woodenwedges. After removing the insulation blocks,the surface was prepared by cleaning withcotton rags and applying glycol to melt andremove any ice formed. The surface was coatedwith a layer of bitumen and the wooden wedgeswere installed immediately. Core samples ofsecond row insulation blocks were collected atseveral locations. The samples were dry,indicating the soundness of the second row ofblocks, which justified the decision to replaceonly the insulation blocks in the outer row (referto Figures 5 and 6). UHDE had alsorecommended replacing only the outer rowinsulation blocks as the inner row blocks werein good condition.

Figure-5 : Core Sample, Second Row

Figure-6 : Core Locations

After supporting the tank, UHDE agreed toreducing the level in the tank further andproceeding to the subsequent decommissioningsteps. The replacement of insulation blockswith wooden wedges in the rest of the outer rowcontinued and was completed in 30 days (referto Figures 7 and 8).

Figure-7 : Wooden Wedges

Figure-8 : Wooden Wedges in place

7.3 Emptying the tank:

The knock-out drum which had been installedbetween the tank and the temporary pump wastaken into service. The pump was connected tothe knock-out drum. The pumping rate wasmonitored hourly and it averaged about25MT/hour. The knock-out drum pressure was10 mbarg which was the compressor suctionheader pressure. The pressure in T-7101B wasaround 55 mbarg. The differential pressure


conditions between the main tank and theknock-out drum remained throughout the levelreduction. During the last stages of pumping,liquid ammonia samples were collected andanalysed for mercury. No mercury was foundpresent. The activities for mitigating the effectsof mercury were, therefore, not considered,although these had been planned.

7.4 Purging the tank:

Once it became impossible to pump theremaining liquid, T-7101B was ready forheating up with hot gas. This step was requiredto evaporate any liquid ammonia in the tanknear the walls in the inner and outer shells aswell as in the pipes in the outer shell. Theammonia tanks were built with a provision to

supply nitrogen to the outer shell and the innershell through a common line. This line suppliesdistributors which are located at floor level. Hotammonia gas from the compressor dischargewas introduced to the nitrogen ring header forheating. The tank has 16 skin thermocouples onthe inner and outer tank shells. Theseindications were provided with appropriatetrending facility in a small control panel. Theheating rate was slow as the temperature of theinlet gas was 20 oC. A small jacketed steamheater was connected in the system andtemperature of the ammonia gas was increasedto 70 oC.

Gradually all the skin temperatures exceeded10oC and heating was considered as complete.

Chart-1 : Heating Profile

During heating of the tank, the pressure wasmaintained at 60 mbarg and vapours wererouted to the refrigeration system. The heatingtook 240 hours (refer to Chart 1).

After completing the heating operation, purgingwas started by admitting nitrogen at tank floorlevel at a rate of flow of 600 Nm3/hr. The tankpressure was maintained at 60 mbarg. Nitrogen

Hot gas purging

-40

-30

-20

-10

0

10

20

Date 28/Apr/04

01:45

29/Apr/04

13:55

01/May/04

02:05

02/May/04

14:15

04/May/04

02:26

05/May/04

14:39

07/May/04

02:52

Time

TE-71116 TE-71117 TE-71118

Total duration ~ 255 hours

°C


was routed through the same jacketed steamheater and the temperature was 70 oC. Thevapours from the top of the tank were analysedfor ammonia content. Inert gas levels in therefrigeration system were monitored based on

the discharge pressure and the purge valveopening. After 3 days, the purge valve openingwas at maximum and the nitrogen analysis at thetop of the tank was 9 % (Refer to Chart-2).

Chart-2 : Nitrogen Purging

Because of high inert levels, it was decided todivert the vapours to the flare. Steam to thenitrogen jacket heater was isolated when thetank base temperature reached 20 oC. This wasnecessary in order to prevent the tanktemperature exceeding ambient temperature.However, the temperatures stabilised at 40 oCwhich is a typical ambient temperature likely toprevail in the Arabian Gulf during the summermonths.

The vapours were analysed for ammonia contentat several points and after 9 days of purging, theconcentration at the top of the tank was 1.58 %ammonia. One of the tank’s safety valves wasremoved and a temporary vent pipe wasinstalled to facilitate local venting. After

confirming the purge gas analysis, nitrogen wasclosed and the tank was depressurised. Themercury content in the vapour was found to be20 micro gram/m3.

7.5 Air purging:

After depressurising the tank, blinds wereinstalled at all tank flanges. The blinds werefabricated of stainless steel material andthickness was chosen to withstand the designline pressure. Stainless steel material waspreferred as it was considered suitable towithstand against cold ammonia which is likelyto be present should any tank side valves pass.The 24” (600mm) liquid line was removed andan air-driven reaction fan was fitted in the

N2 Purging

0

10

20

30

40

50

60

70

80

90

100

9-May-04 11-May-04 13-May-04 15-May-04 17-May-04 19-May-04 21-May-04

NH3 analysis

NH3 Concentration (%)


nozzle. The top central as well as peripherymanways were opened and sweetening wasstarted. A manometer was installed at the tankbottom to closely monitor the tank’s internalpressure at low ranges because of the highcapacity of the reaction fan. The tank safetyvalves and vacuum relief valves were removedand electric air movers were mounted on thesenozzles.The outer tank manway cover was removed andafter 3 days of sweetening, the manwaydiaphragm plate was cut using the cold cuttingtechnique (refer to Figure 9).

Figure-9 : Cold Cutting the Diaphragm Plate

The ammonia content in the annular space was12 ppm and mercury was 5 micro gram/m3

maximum. The inner tank manway cover wasopened using full Personal ProtectiveEquipment (PPE). The same cold cuttingtechnique was adopted for removing inner tankmanway diaphragm plate.

7.6 Tank entry and cleaning:

The tank was ready for entry after 40 days.Initial entry was made using full PPE toascertain the condition inside the tank.Ammonia concentration around the floor was 10ppm and mercury was 8 micro gram/m3, both ofwhich are well below the allowable limits. Theallowable exposure limit for mercury is 100micro gram/m3 (OSHA standard). Oxygenlevels were normal at 20.8%.The first inspection of the tank revealed a finefilm of oil over the entire floor and small

puddles in a number of places due to theunevenness of the floor plates. Oil hadcollected against the outer periphery plate for adepth of maximum 20 mm since the tank floorplates have a fall of 25 mm from the centretowards the circumference. A two-stagedapproach was adopted for removing oil. First,the bulk oil was removed using vacuum truckand dispatched to a dedicated disposal facility.A total of 5 m3 of oil was collected from thetank (refer to Fig. 10).

Figure-10 : Removing Oil

After removing the oil, the surface still had a lotof oil residue, puddles as well as an oil coatingover the wall plates at the bottom remained. Itwas decided to clean the floor plates includingthe tank wall surfaces up to the first course weldjoint using a high pressure hydrojet. Whilehydrojet cleaning, the oily water was removedcontinuously using the vacuum truck (refer toFigure-11).

Figure-11 : Hydrojetting the Tank floor


A total of 35 m3 oil / water mixture was removedand stored in dedicated tanks. The hydrojetcleaning took 5 days. The required 50V lightingwith centrally earthed transformers, were providedand power cables were inserted from the topthrough one of the safety valve nozzles for safety.The bottom manways were kept free of cables andhoses to ensure quick evacuation of personnel, ifrequired. The utility hoses were routed throughthe 24” (600mm) liquid discharge line into thetank. After hydrojet cleaning, the floor surfaceand wall became oil free and dry and posed noslipping hazard.

The oily water mixture was analysed to contain7 ppm of oil only and could therefore bedischarged safely to the holding pit of theplant’s water treatment unit for further disposal.The entire cleaning activity took 12 days.

8.0 Tank Inspection

8.1 T-7101B – Inspection Scope:

In line with the EFMA guidelines, the scope ofinspection had been finalised in the projectformulation stage. The detailed scope is asfollows (refer Schematic-2):

Schematic-2 : Inspection Scope

8.2 Preparations for inspection:

Mobile scaffolding was erected for the internaltank shell inspection of the first and secondcourse weld joints. For the annular shell firstcourse weld joints, a small wooden structurewas suitably arranged in line with the curvature

of the annular space. A fixed scaffolding towerwas installed covering the required EFMAscope of 10% T joints up to the top of the innertank (refer to Figure-12).

Rope access methods were planned for theinternal roof surface and annular space between

Floor – Inner Tank &Annular Space100% MPI100% Vacuum Box testUT Thickness Scan

Inner Tank Inner Wall100% FMPI (1st & 2ndcourse & 10% T-Joints ofother courses)UT Thickness Scanning

Welds for ladders*, sumps,wall stiffeners* etc.100% NDT

Annular Rings Welding*random NDT of welds

Inner Tank Outer SurfaceOuter Tank Inner Surface100% FMPI of 1st course

Pipe Supports*100% NDT(Rope Access)

Anchors 65 Nos.100% FMPI

Anti-rotation Device 8 Nos.100% FMPI

Bottom plateVisual Inspection afterremoval of insulationbricks

External:100% leak test100% Vacuum Box testRandom UT ThicknessScan(4-6 points / plate)Internal*:100% NDT of nozzlewelds & shell to roofjoint

RoofGeneral:Nozzles 29 No. (roof &shell) 100%NDTMetallurgical Replication/ hardness etc. – Needbasis* Rope Access


the two tanks. The required scaffolding wasbrought into the tank through the peripheral roofmanway. All the weld joints were power-brushed to requirement of the InspectionSection.

Figure-12 : Fixed Scaffolding Tower

Extensive safety checks and preparations weremade for entry to the tank.

- A Confined Space Rescue Plan wasprepared and provided to the team.

- A Confined Space Entry Control Formatwas developed and administered by adedicated Safety Operator. This included aboard displaying the company badges of thepersonnel inside the tank.

- A suitable warning arrangement wasinstalled inside the tank to alert personnel incase of emergency in the complex.

- Air Chillers were provided for the comfortof the personnel as hot summer weatherconditions prevailed.

- One reaction fan was kept running at the topof the tank to provide ventilation.

- All personnel entering the tank wereprovided with torch lights as a protectionagainst power failure.

- An Emergency Evacuation exercise wascarried out to check the adequacy ofpreparations and capabilities of the overallsystem.

- An independent Safety Auditor wasappointed to monitor all work permits andrelated preparations.

8.3 Inspection Findings:

1) Inner Tank – Shell:

All the cleat marks on the first and second shellcourses were cleaned by buffing and slightgrinding. Fluorescent Magnetic Particleinvestigation (FMPI) was carried out on all cleatmarks. A total of 23 linear indications wereobserved mainly on the first course shell (referto Figure-13).

Figure-13 : Example of Linear Indication

All defects were removed by grinding. Thedepth of defects varied from 1.0 mm to 1.5 mm.FMPI was carried out on all locations aftergrinding and all were found to be satisfactorywith no recordable indications (refer to Figure-14).

Figure-14 : FMPI in progress


On-site metallurgical examinations were alsocarried out on 10 selected locations where linearindications were identified by FMPI. Theservices of a metallurgical specialist wereengaged for this task. The metallurgical studyrevealed that all the indications examined, bothexternally and internally, were original defectscaused by either fabrication (lack of fusion,overlap or hydrogen cracking) or minorlaminations formed during the plate rollingprocess. The metallurgical specialist studied theselected locations by replica test methods andconcluded that these were not stress corrosioncracking marks (refer to Figure-15).

Figure-15 : Metallurgical Testing

2) Inner Tank – Bottom Plate:

FMPI was carried out on the bottom plate weldsand these were found normal except for onelinear indication of 10 mm long on the annularplate (refer to Figure-16). The defect wasremoved by grinding and FMPI revealed norecordable indications afterwards.

Figure-16 : Floor Plate Inspection

Vacuum box leakage tests were carried out onall bottom plate welds. No leakage was detected(refer to Figure-17). Ultrasonic thicknessmeasurements were carried out and the resultswere normal.

Figure-17 : Vacuum Box Leakage Test

3) Outer Tank:

FMPI was carried out on all 65 external tanksecuring anchor straps, shell to annular platewelds and tank nozzles (refer to Figure-18).

Figure-18 : FMPI Work on Outer Tank

Twelve (12) anchor straps were found to havelinear indications, classified as originalconstruction defects. All indications wereground to a depth of 1 mm and further FMPIrevealed complete removal of these indications.FMPI was done for annular welds (betweeninner and outer shell) and no defects werefound.


4) Roof:

Externally, an ultrasonic thickness survey wasdone for all the roof plates. Vacuum box testswere performed for all roof plate welds jointsand FMPI testing was carried out for all nozzlesand structural attachment to the roof.

Internally, eddy current examinations werecarried out on all central nozzle welds and thecircumferential weld at the roof to outer shelljoint. The under-side of the roof was visuallyinspected and the condition of the epoxy coatingwas found to be satisfactory. The internalinspection of the roof was done utilising ropeaccess techniques (refer to Figure-19).

Figure-19 : Rope Access Technique

9.0 Recommissioning of TankT-7101B

After the successful completion of theinspection and other related modification work,the recommissioning program below wasfollowed.

9.1 Leak test with air/integrity checks:

After boxing up the manways, the tank pressurewas gradually increased to 70 mbarg byintroducing air from a portable compressor.Leak checks of all flanges were performed andsubsequently the pressure was slowly increasedto 100 mbarg. The pressure was held for 30minutes and all the nozzles, anchor straps,foundations, etc. were checked for integrity andfound normal. After that the tank wasdepressurised to proceed with nitrogen purging.

9.2 Nitrogen Purging:

The purging with nitrogen started at a rate of200 Nm3/ hr., vented at the top of the tank. Anitrogen shortage at the time enforced a lowerrate of purging.

After 15 days, the Oxygen level in the tank wasfound to be 2%. Removal of all blinds wasstarted after reducing the pressure to 5 mbargand maintaining nitrogen flow at 100 Nm3/hr.After removal of the blinds, the tank pressurewas again increased to 70 mbarg and all flangesconnected to the tank were again leak tested.

It was decided to reduce the Oxygen level toless than 0.2% since the presence of Oxygen,even in very minor concentrations in liquefiedammonia, can lead to a considerably increasedrisk of Stress Corrosion Cracking (SCC). Toachieve these very low concentrations, apressurising and depressurising technique wasadopted. These step actions effectively strip outpockets of Oxygen which are likely to exist inthe tank under continuous purging conditionsbecause of stratification.

The Oxygen level in the tank was reduced to0.15%, which was confirmed by four repeatedsamples. The duration of purging was 20 days,a longer period than expected because of thenitrogen shortage mentioned (refer to Chart-3).


0

1

2

3

4

5

6

7

6-Aug-04 11-Aug-04 16-Aug-04 21-Aug-04 26-Aug-04 31-Aug-04

O2 c

once

ntr

atio

n % Top PI

Top ventOuter shell

Chart-3 : Oxygen Levels

9.3 Water addition to the tank:

Based on UHDE’s recommendation, water wasadded to avoid direct impingement of liquefiedammonia droplets on the bottom plate duringthe cooling down period. These may createlocalised stress and possible cracks. The pool ofwater on the bottom plate ensured a uniformcooling effect over the entire bottom plate. Inaccordance with the dimensional calculations avolume of 20 m3 of demineralised water wasintroduced to the tank for this purpose.

9.4 Ammonia-Nitrogen exchange:

It was planned to add ammonia vapour to thetop of the tank and to withdraw the ammonia-nitrogen vapours from the bottom, followed byflaring. This exchange operation was completedin 5 days. Subsequently the tank vapour outlet

was connected to the suction of the refrigerationcompressors and arrangements were made tostart cooling down operations.

9.5 Tank cooling down:

To facilitate cooling down, liquid ammonia wasintroduced from the top of the tank. It wasenvisaged to utilise the loading line as liquidbuffer from where the liquid ammonia wasintroduced to the central distributor at the top ofthe tank, using the cool down line. The totalexpected cooling down time was 54 hours,based on a 1 oC/hr cooling rate, however forsafety reasons the cooling period was doubledusing a cooling rate of 0.5 oC/hr. Cooling wasclosely monitored using the bottom plate andshell skin temperatures. The tank bottom platetemperature reached -31 oC after 7 days (refer toChart-4).


Chart-4 : Cooling Rate

Subsequently, liquid ammonia at a rate of 2MT/hr was introduced to the tank until levelindication was obtained. Later the tank levelwas increased to 300 mm to enable the tank andinsulation system temperature to reach anequilibrium profile.The tank was recommissioned successfully on14/09/2004 and is in operation in a satisfactorymanner since that time.

10.0 Inspection of Tank T-7101A

Based on the experience of inspecting T-7101B,GPIC has successfully completed the inspection

and recommissioning of the other ammoniastorage tank, T-7101A, during the first quarterof 2006. The same methodology and procedureswere adopted for T-7101A. The temporarypiping and relevant decommissioningarrangements were adapted and relocated to suitT-7101A. The complete project, includingsupervision of the insulation block replacement,was executed by the GPIC team. Roofinsulation was not replaced as it was in goodcondition. Salient features of the inspection ofT-7101A, along with T-7101B, are tabulatedbelow in Table-3.

T-7101 B COOLING DOWN(5HRS AVG .)

-35

-30

-25

-20

-15

-10

-5

0

5

10

15

20

25

30

35

40

8-Sep 9-Sep 10-Sep 11-Sep 12-Sep 13-Sep 14-Sep 15-Sep

TEMP. DEG. C

TE-71116

TE-71117

TE-71118

TE-71119

TE-71120

TE-71121

TE-71122

TE-71123

TE-71124

TE-71125

TE-71126

TE-71127

TE-71129

TE-71130

TE-71131

0 HRS 82 HRS58 HRSHRS

10 HRS 34 HRS 106 HRS 134 HRS


Description T-7101B T-7101A

1) Year of decommissioning 2004 2006

2) Decommissioning method Dry method with hot vapour addition. Dry method with hot vapour addition.

3) Jobs carried out - Outer row insulation blocks replacement.- Roof insulation renewal.- 6” & 24” liquid O/L bellows change.- Enraf level monitor modifications.

- Outer row insulation blocks replacement.- 6” & 24” liquid O/L bellows change.- Enraf level monitor modifications.

4) Salient Inspection findings:

(a) Inner tank shell - 26 Nos. linear indications with max. 1.5 mm. (at cleat areas).

- 4 Nos. linear indications with max. 10 mm.(at cleat areas).

(b) Bottom plate - 1 linear indication. - No linear indication.

(c) Outer tank - Linear indications at 12 anchors. - Linear indications at 9 anchors.

(d) Roof - Normal. - Normal.

(e) Metallurgical Study - Linear indications of shell cleat areas revealeddefects due to Hydrogen crack/lack of fusion.

- Fabrication defects (removed by grinding).

- All linear indications were due to fabricationdefects and no evidence of cracks due tofatigue or SCC.

- No evidence of service related crack growthexternally (due to fatigue) or internally (due toStress Corrosion Cracking).

(f) Repairs carried out None. None.

5) Project duration: Days Days

(a) Wooden wedging (initial 65 locations) 10 7

(b) Wooden wedging (remaining locations) 30 20

(c) Tank emptying 5 5

(d) Hot gas purging 10 10

(e) N2 purging 15 15

(f) Tank cleaning 12 12

(g) Inspection 20 15

(h) Recommissioning 45 30

TOTAL NO. OF DAYS 147 114

Table-3 : Salient Features of both ammonia Storage Tanks inspected

11.0 Difficulties Faced

11.1 Ammonia released from sand screed:

A 50 mm deep cement-sand screed ring isprovided between the inner tank cup bottom andthe outer tank bottom. The width of the ringlayer is 900 mm and it protrudes by 125 mmoutside the inner shell into the annular space.

Longitudinal cracks were observed in theprotruding layer and these were repaired. Evenafter emptying and purging the tank, ammoniawas released from the sand screed and it wasdifficult to work in the annular space (refer toFigure-20). The annular space inspection andsand screed repair work were delayed in boththe tanks due to the presence of this ammonia.Work could only be completed after applyingadequate safety measures.


Figure-20 : Annular Space Arrangement

12.0 Improvements

12.1 Provision of a skirt and insulation forthe tank foundation:

To prevent ingress of water into the bottominsulation blocks through the foundation plinth askirting cover was provided for the fullcircumference (refer to Figure-21).However, the provision of a skirting covercannot eliminate condensation on thefoundation. Water vapour inside the void spacebetween skirt and tank wall can condense on thefoundation as a result of the low temperaturesthere.

Figure-21 : Skirt Arrangement


The void space between the skirt and the tankwas filled with polyurethane foam to provide avapour barrier to avoid condensation (refer toFigure-22).

Figure-22 : Skirt Insulation

13.0 Conclusion

(a) The successful inspection of the ammoniastorage tanks at GPIC ensured the following:

• Improved safety and reliability of theammonia storage tanks T-7101A and B.

• Assurance of tank integrity.• Improved integrity of the tank

foundations as a result of thereplacement of damaged insulationblocks with wooden wedges.

(b) After 2 years of operation, the woodenwedge blocks in the foundation of T-7101Bwere checked and observed to be inexcellent condition with no icing. Thissituation more than justified the decision toreplace the insulation blocks with woodenwedges. (refer to Figure-23).

Figure-23 : Wooden Blocks Inspection

(c) Both the inspection projects wereaccomplished safely without any incident.

(d) The detailed schedule, operating proceduresand comprehensive checklists ensured theprogress of the inspection work inaccordance with the plan. The effectiveinterdepartmental communications andregular reviews contributed much to theoverall success of these critical projects.


Documents

Successful inspection of two large Ammonia Storage Tanks · Ltd. The tanks were commissioned by Snamprogetti in 1985. The storage tanks are cylindrical with domed roofs and designed