You are on page 1of 11

Successful and safe de- and recommissioning

of a cold ammonia storage tank

The paper that Continental Engineers BV and OCP present discusses the decommissioning procedure
to carefully and safely empty and inertize a cold ammonia storage tank and the problems that could and have occurred.
The decommissioning took place in April/May/June 2004 and all together it covered a period of 6 weeks (emptying 4 weeks, inertizing 2 weeks).
It also discusses the recommissioning of the same tank, which took place several months later.
For the recommissioning first the oxygen had to be removed until a level of <1.5 volume % had been reached and after
that a controlled filling with gaseous ammonia should take place in order to avoid too cold spots in the steel bottom or
wall which could cause stress cracks. Recommissioning took a period of 2,5 weeks and went very successfully thanks
to the excellent preparation and help of OCP management and employees.
The tank was cleared and finally accepted by OCP on 19 October 2004 as being successfully recommissioned.

L.A.J. Tol and G.J. Tol

Continental Engineers BV - The Netherlands
in cooperation with:
A. Bourras
OCP Jorf Lasfar at El Jadida, Morocco





sioning of the cold ammonia storage tank IR5502. After

that, inspection and re-insulation should take place and
the tank would have to be recommissioned, i.e. safely
refilled with ammonia until the fluid phase of 33oC
was reached.


ffice Cherifien des Phosphates Group

(OCP) was founded in 1920 and its
achievements are:
75 % of world phosphate reserves (98% in the
middle of the country, 2% in the south part)
85 % of the world phosphates production is
intended for fertilizer production.

In March 2004 Continental Engineers

BV (C E) provided OCP with the required procedures
for the de- and recommissioning of the mentioned ammoniastoragetank nr. IR5502, of which some data are
given below:
Liquid ammonia:
Temperature of liquid:
Height of wall:
Material of wall:
Height of dome:
Material of dome:
Design code

OCPs production sites, production levels and position are as follows:

1. Production sites
4 phosphate mines in the country
(Khouribga, Bengurir, Youssoufia
and Boucra/Layoune).
2 coastal sites of chemical transformation (Safi and Jorf Lasfar)
4 shipment ports (Casablanca, Jorf
Lasfar, Safi and Layoune)

ASTM A516-70
API 620 app.

23,000 m3
120 mm (polyurethane).
Annular space between tank wall and concrete out
side wall: 1.20 m

Production levels:
Phosphate production: 23 Mt (52% are
transformed in phosphoric acid and
- Phosphates:
12.0 Mt
- Phosphoric acid:
1.6 Mt P2O5
- Fertilizers:
2.2 Mt

The tank is placed on a concrete ring foundation which

is protected against freezing by a system of heating
coils. The center of the ring is filled up with a layer of
sand and a layer of 200 mm foam glass.

3. Position in the world market:

Leader in phosphates exportation under all its forms
Leader in phosphoric acid exportation


Preparing the Procedure:


Preparatory observations and actions:

Before the tank can be inspected, it has to be emptied and all liquid and gaseous ammonia would have to
be removed from the tank. The pump suction line was
however located somewhat above the bottom of the
tank (300 mm) which means that a certain amount of
liquid ammonia remains in the tank, amounting to, by
estimation, approx. 300 t. It is important to state that
there were no manholes available and there was not a
provision for draining below tank bottom level. See fig.

To produce their fertilizer, OCP imports ammonia

(NH3) by ship in the total amount of 400,000 t/a.
A great deal of this ammonia is transported to
OCPs site of Jorf Lasfar at El Jadida, a place 100 km
from Casablanca and located near the Atlantic coast and
stored in two big tanks with a volume of 23,000 m3.
OCP wanted to overhaul one of these tanks in 2004
(for the first time since the tank was put into operation
in 1987) and they therefore inquired early 2004 at Continental Engineers BV (C E) at Zaandam, The Netherlands, as reputed ammonia-engineers whether they
could prepare a solid procedure for the safe decommis-


-33 oC
35 m
24 m

The route to use aqueous ammonia or to flare was

not feasible at OCPs location which meant that all re-



maining ammonia would have to be removed by evaporation and drawn off as vapour.
Purging of the tank with nitrogen prior to entrance
of oxygen (air) was necessary to avoid the (explosion)
risk of ammonia in an oxygen-rich environment and
which could occur in case of a 15-28 vol-% ammonia
concentration. However, to minimise eventually the required heat for evaporation one could first reduce the
liquid level in the tank by increasing the operating pressure of the tank (see below).

Practicle calculations
Increasing the pressure by increasing the set point
of the compressor, taking into account that the pressure
relief valves at the tank were set at 1.16 bara. According to our calculations indicating a pressure increase of
0.03 bar, which was fully acceptable, the pump could
remove an extra 200 t. of liquid ammonia at reduced
flow and the required time for that should be approx. 50
See fig. 2 with schematic diagram.

Possible options to remove the remaining ammonia

as considered by C E are:

A quantity of 100 t ammonia should then remain in the

tank and after that we could pump out another part of
the ammonia by means of a flexible hose which could
result in an extra reduction of at least 50 t. The left over
then would have to be evaporated
For each of the aforementioned options, calculations were made and the required decommissioning
time was estimated resulting in:

1. A combination of pumping liquid ammonia out

of the tank through a flexible hose via the
suctionline and evaporating the remaining liquid with hot ammonia vapour;
2. Using only hot ammonia vapour;
3. Natural evaporation;
4. Adding water to the tank and pumping aqueous
ammonia out of the tank;
5. Evaporating the remaining liquid by heating
annular space between the concrete wall and
the storage tank (in that case the insulation had
to be removed).

Option no.:

Estimated time/period for only empty-

NB: 150 kg/h

approx. calculated quantity of vaporized ammonia

1. 12 days (pumping 60 t. out in 1 day and 11

for evaporation of 40 t.)
2. 27 days (100 t. / 150 kg/h)
3. 16-32 days (boiling off rate 0.02%-0.04%,
dependant of insulation quality of tank and
ambient conditions)
4. not applicable nor feasible at OCPs (see also
point 2.1)
5. 10 days, however not practible nor feasible

Be well prepared before you start!:

Necessary equipment & tools

Before entering into the right procedure and

executing the necessary activities, certain provisions have to be made such as:
a storage facility for liquid ammonia,
availability of hot gaseous ammonia (45
availability of nitrogen for purging purposes,
flexible hoses and suitable pumps, (spare)
heater(s) for superheating ammoniavapour,
a fan for ventilation purposes etc. etc.

Conclusions on methods for emptying:

Execution by method 1 was considered not
to be safe, because of our (previous) bad
experience with bringing in flexible hoses
Methods 2 and 3 were considered as feasible methods on OCPs site
Method 4: not applicable because of lack
of enough water and OCP was not in a position to use aqueous ammonia
Method 5: not feasible because of safety
but also for practicable reasons
Based on the required time, the required investments and, most of all, practical considerations C E advised OCP to use method 3 natural evaporation for
emptying the tank.




2.3 Purging requirements:

Decommissioning and encountered problems,

After all the liquid NH3 is evaporated, purging with

N2 of the tank can be started. For this purging the availability of an amount of 30,000 m3 of nitrogen was required. The nitrogen should have to be carefully inserted since the layers of ammonia and nitrogen clouds
can mix up with eachother when this is done "too wild";
this because of their different densities. While pushing
out the ammonia the concentration of it should be
measured at the outlet of the tank towards the refrigeration compressor. Assuming a low line velocity of 3 m/s
through a DN250 line the expected purging time should
amount to 60-70 hrs (< 4 days).

or in other words: Reality Check of Practice vs. Theory

The assumption was that 100 t. of ammonia should remain in the tank after pumping out a part by means of
pressure increase in the tank. However our commissioning manager could not check nor verify the level of
ammonia in the tank. OCP could also not give a reliable
measurement of the amount in the tank, so the first uncertainty arose.
The measured levels varied from 28 cm to 45 cm liquid
and the question was:
Is the content 100 t. [~145 m3] or is it (much) more
than the measurements indicated
(250 t.= 38 cm height)? A lot of time was spent to install a reliable measuring device by means of a floater, a
thick shelf and finally steel buses. This was all done
while not in operation.

2.4 Air ventilation before inspection of the tank:

Before inspection by maintenance people could
take place, ventilation by air has to be applied. This
could be performed by means of a high perfomance fan
of 50,000 Nm3/h blowing through a manhole.
The nitrogen content has to be measured and when the
O2-concentration is in the range of 20 to 21% ventilation can be stopped and maintenance and inspection
personnel can then safely enter the tank.
The required time for air ventilation based on comparable situations was estimated at 7 days.
In total the required, estimated time period for emptying, nitrogen purging and air ventilation would cover
probably 30 days.

Furthermore, despite the advice of C E to use option 3,

but mainly because of the assumed higher content of
ammonia, it was decided to blow in hot ammonia in
order to speed up the evaporation rate. Flexible hoses
were brought in into the tank, one through the underpressure relief valve and the other through the overpressure relief valve. However, one of the flexible hoses
brought in from the top was too short.
On the other hand emptying the liquid out of the
tank by using the pumps was not successful since we
did not get out the expected amount possibly due to

2.5 Permission to start:

Parallelly it was locally investigated whether one

could use the watermethod described under option 4,
in order to come to a 2 % solution of ammonia and water for disposal off to the sewer of the harbour, but this
meant that approx. 7500 m3 of (demi)water should be
needed. A 25% aqueous ammonia could however not
be handled by OCP and so both ideas had to be abandonned.
Because of the forced methods to at first remove
a big amount of ammonia would not fulfil the requirements to empty safely and adequately the tank, it was
decided to go back to the original method, i.e. emptying
by natural evaporation.

Continental Engineers BV produced based on its

experience, the facts and the figures, a detailed procedure for the overall decommissioning of the ammonia
storage tank. This procedure was presented to OCPs
management in both English and French language.
Based on this extensive procedure, OCP ordered C E to
start with the proposed activities mid April 2004.
See fig. 5 for overall schematic diagram of the ammonia storage system for cooling, loading and emptying




The expected rate of the liquid level reduction was

approx. 1 cm/day equal to 6.5 t ammonia/day, but in
practice it did not work that way. Perhaps the tank was
too well insulated.
To improve the evaporation rate of ammonia
OCP removed the insulation at several places (which
had to be removed anyway for renewal) by taking out
2x2 m spots allowing entering of heat through the wall.

6. Owner should, at least for future to build tanks,

make a drain beneath the tank bottom level to
secure emptying the liquid from the tank by
means of pumps.
7. Owner should install accessible manholes in
the tank for inspection, maintenance and for
measuring purposes.
3.2 Nitrogen purge & Air Purge

The first conclusions of the reality check were:

The pumps did for some unsolved reasons not
fulfil the expected duty (possibly gassing)
We could not check, by removing parts of the
insulation shield, from the outside of the wall
what the level was (one could not observe a
distinct colour difference)
There was not 100 t of ammonia in the tank but
(much) more (some 250 t)
Flexible hoses should be available and if yes
they should be of enough length; it is however
sometime quite difficult to insert these properly
into the tank and liquid.
One could not adequately measure the liquid
level i.e. the remaining quantity of ammonia
The alternative methods could not be applied
or did not work
After all this, the schedule (30 days) could not
be maintained (i.e. more time was required, in
this case ~45 days).

Purging with nitrogen (N2) is based on the principle

of creating a difference of density in the layers of NH3
(molecular weight 17) and nitrogen (molecular weight
28). For environmental reasons the tank is connected to
the other tank because of pushing out the NH3.
This can be achieved by adding nitrogen very
slowly (1.5 m/s, DN 250) in order to avoid turbulence
via the manchet in the suction line to the bottom of the
tank and take off the ammonia vapor with the compressors. When this is done very carefully, there is a sharp
separation between the two cloud layers. When the
level of nitrogen is about 1 meter above tank bottom,
the amount of nitrogen can be increased (v=3 m/s), and
when the level is about 2 meters above the tank bottom,
the amount can be increased to maximum capacity
(~800 m3/h).
One should monitor the pressure in the tank (which
in the OCP case has to stay below 100 mbar).
We then could calculate how many hours it should
take to reach the top of the tank (24 m). We needed an
amount of approx. 30,000 m3 of nitrogen and it took
eventually 2 days to feed this amount (in stead of 4).
When this level was reached, the two tanks could be
separated and the mixture of ammonia and the last nitrogen could be blown through a water tank in order to
avoid environmental hazard.

Lessons learnt for both parties at the phase

of emptying the tank as part of the
overall decommissioning:

1. Before preparing a (desk) procedure the engineer should execute a site visit to make notice
of local circumstances and discuss the planned
2. It is advisable to make a very practicable procedure (with step-by-step actions and explanatory drawings) in stead of a desk report with
theoretical assumptions and schemes.
3. The engineer should verify and list all the necessary measuring devices, auxiliary tools &
equipment and ensure the local availability.
4. Looking back the method with inserting hot
gases had to be preferred.
5. Contingency (x %) for the required time period
should be timely communicated and built in to
the proposal towards client/owner.


After two days already we could install the air fans [ ]

for the purge with air via the manholes on the top and
via the suction line at the bottom, in order to ensure a
safe environment for maintenance and inspection personnel.
See fig. 3 for a schematical diagram of air ventilation
On the top of the tank we installed oxygen detectors
to measure O2-content of the gas/air mixture. When the
oxygen level is approx. 20 % (to be measured with
Drger tubes or similar) and the NH3-concentration is <
20 ppm (MAC value) the environment is safe for entering the tank. The air purge period took approx. 3 days.



Before entering the tank first of all enough lighting

should be brought in and men should first inspect
whether there was oil or other dirt on the bottom, which
should be removed first.
Apart from the measurement by means of official
detection the rabbit proof was also applied by the client.
This is an almost fail safe test to check whether a man
can safely enter such a tank
The method is simple: catch a rabbit and let it in into
the tank and look if it survives after a while so:
See fig. 4:
Schematic drawing of a successful rabbit proof.


First of all, all the air in the tank should be evacuated by feeding nitrogen into the tank, but why?
It is to state that the presence of oxygen in connection
with (liquid) ammonia can lead to a considerable increased risk of stresss corrosion cracking (SCC) of
tank material. Another risk is the build up of electrical
charge due to static electricity while spraying liquid
ammonia into an oxygen containing tank. The explosion risk will be there when the air contains between 15
and 28 % of ammonia gases.
So, first the air will be replaced by nitrogen and
just after that gaseous ammonia, and later liquid ammonia, can be brought into the tank. During this process
the tank temperature will be decreasing.
This has to be done safely and slowly in order to let
the tank follow the required temperature change of
less than or equal to 1 oC/h.
When nitrogen is inserted the oxygen content of the
outcoming air will be continuously measured and one
can see when the atmosphere in the tank becomes inert. C E estimated that this should take a few days and
actually after 4 days the O2-level in the tank was <1.5
%: we actually measured 1.14% O2 and this is considered to be a sufficiently safe value to start with blowing in gaseous ammonia..
At the same time the tank had to be slightly pressurized
to approx. 1,15 bara in order to observe and check
whether all the connections are tight (by using bells of
soap) and which took an extra day.

It was agreed between parties that when the rabbit

survived at least 15 min. the environment was to be
considered safe for mankind. While the detectionmeters
at top, bottom and manholes indicated an oxygen level
between 20.3 and 20.7 %, which was o.k., the rabbit indeed survived this exercise.
The client undersigned mid June 2004 the taking
over protocol stating that the decommissioning, however with some delay, was finalised. In stead of the estimated 4 weeks the total decommissioning period took
2 weeks extra, but at least everything went in a controlled and a safe way.


Inspection and re-insulation

OCP executed after the decommissioning the inspection of the tank which gave no major problems for
the maintenance department. The re-insulation was executed by a Morroccan company and the whole wall and
dome , approx. 4000 m2 , were covered with a new layer
of 120 mm polyurethane.


Nitrogen purging and safety measures to

avoid the risks

However, to execute this nitrogen purge, one should

take into account a few safety measures up front:
All inspections and repairs should have taken
place according to rules and legislation;
All manholes shall be closed;
All relief valves to be checked and installed;
Availability of (enough) nitrogen for purging ;
Availability of (enough) liquid ammonia for
filling and cooling the tank gradually.

Recommissioning and its features

In September 2004 OCP was ready with the reinsulation and was ready to recommission the tank i.e.
it could be filled with ammonia gas which should
slowly be turned into cold liquid ammonia at 33 oC.


Bringing in ammonia vapour into the

The procedure of feeding the ammonia vapour into the

tank (at a rate of 500 Nm3/h, temp. ~40 oC)) should take
several days and each time the laboratory (of the client)
has to measure the ammonia content in the tank. The
ammonia gas was taken from the other storage tank [nr.




1] on the premises of OCP and the liquid ammonia was

later taken from the sphere.
The actual NH3-values in the tank were measured
each day:
After one day the value was: 14 gr/m3 (~2 %)
After two days it was:
38 gr/m3 (~5.5 %)
At the end after three days the value measured was:

5.4 Closing of the recommissioning

Due to excellent teamwork of all people involved, the
recommissioning went very well and in total it covered
approx. a time period of 18 days. No major problems
occurred during this phase except for the blow off of a
relief valve of a heat exchanger vessel.
We can conclude that this recommissioning was indeed safely and successfully executed.

87 gr/m3 (~12.5 %)
One should take samples from the tank until the N2content was below 5%; this is a measure to start with
bringing in liquid ammonia for filling and further cooling down of the tank.
No major incidents or problems occurred during this
phase of recommissioning; only one relief valve from a
HE-vessel opened and blew off above the annular
space. Because of the well taken safety precautions, i.e.
no persons were allowed by OCP in the neighbourhood,
there was only waiting time to let the ammonia vapourize. The total ammonia vapour time amounted to
approx. 6.5 days.


Summary and conclusions

Thanks to the excellent cooperation between the commissioning manager of C E, the management and the
operating personnel of OCP on site the decommissioning, although quite some (unforeseen) problems occurred, went well and after that the recommissioning
even went perfectly.
It is of major importance to prepare a procedure,
which is supported by experienced commissioning personnel and which is to be accepted by the clients representatives on forehand.
Also up front the client should take the necessary
measurements on site in order to secure a smooth execution of the activities and OCP did so. There will be
however always (unforeseen) problems on site, but only
by good communication, expert consultation and
enough patience, one can succeed in achieving the set

Cooling down the tank by spraying liquid ammonia.
It is to be advised that the temperature drop has to
be uniformly and be lower than 1 oC/h. The ammonia
will be sprayed in from the top and will liquefy after a
certain time period on the bottom of the tank, subsequentely cooling down this bottom.
At the end of this activity the temperature at the
bottom will be -32 oC, which means that with an ambient temperature of +20 oC, the total difference will be at
least 52 oC. It should therefore require a cooling down
time of at least 52 hours, but for safety reasons we have
chosen 3 x as much, i.e. 156 hours (~6.5 days).
At a flow rate of 70 to 120 kg/h the liquid ammonia
has to brought into the tank; during this procedure the
temperature at the bottom and alongside the wall have
to be carefully monitored by the responsible people. If
the temperature decreases too fast one should immediately slow down or stop the filling procedure to avoid
any risk of stress corrosion cracking of the bottom or
the side walls. It took 4 to 5 days of filling with liquid
ammonia (approx. 20 t.) to cool down the tank to the
level of -32 oC (actually measured).

Continental Engineers likes to thank OCP that they

were trustworthy to give the order for this decommissioning as well as the recommissioning to them. Also
our de/recommissioning manager, Mr. Gerrit Tol has to
be mentioned since he was the solid and cool factor
C E on OCPs site during periods that the inside and
outside temperatures were quite high..
Finally C E like to thank OCP, especially Mr. Abdeljalil
Bourras, manager projects dpt., for giving us the opportunity and his cooperation to prepare and present this
paper for the interested audience here at Toronto.
See fig. 6: Photo of mr. Bourras of OCP (r.) and mr.
L. Tol (l.) of C front of the ammoniatank.





Attachments to paper 523 AIChE Toronto

Page 1

R obe du

n o r m a l m in im u m liq u id le v e l

D N 250
lig n e d e
s o u tir a g e

250 m m

c e n te r lin e

460 m m
300 m m
75 m m
100 m m

Fond du

Fig. 1: Location of suction pipeline near bottom of tank



Tank 1
p=5 mb


pump filter

tank 2
p=150 mb

Fig. 2: Schematic image of increasing the pressure in tank 2




Attachments to paper 523 AIChE Toronto

Page 2

3 fans at middle manhole

one fan at outer manhole

one fan via the suction line

Fig 3.: Extensive air ventilation by means of (5) air fans, exhaust via nozzles
(all valves were removed).

After 15 minutes

Fig. 4:


The rabbit proof for checking safe environment in tank after purging with nitrogen resp. air.







Fig. 5





NH3 gaz
Ref. comp





DN 400 AL









DN 200 AG


Attachments to paper 523 AIChE Toronto

Page 4 (Photograph)

Fig. 6: Photograph in front of the big ammonia storage tank of Mr. A. Bourras of OCP (r.), director of
projects, and the author of the paper, mr. L.A.J. Tol (l.) of Continental Engineers BV, The Netherlands
at the location of El Jadida, Morocco.