CRACK FORMATION IN STORAGE TANKS

Anders Nielsen
Haldor Topsoe
Vedbaek, Denmark

Anhydrous ammonia is widely used as fertilizer in Denmark. Development of testing technique

The quantity in 1970 is expected to be approximately
128,000 metric tons. Essentially all of the anhydrous am- The technique normally applied, as mentioned above, has
monia used in Denmark is distributed by A/S Ammonia. been the magnetic particle method also called the magneto-
Denmark is geographically a country of one peninsula and flux method.
many islands. A widespread system of storage facilities for
anhydrous ammonia is therefore required to serve the Dan- The sensitivity of the magnetic particle method varies
ish farmers. within wide limits and according to the experience gained
only a method giving a high sensitivity should be relied
Until recently it was the common opinion in Denmark upon for this inspection.
that the storage of liquid ammonia in mild steel tanks under
pressure and at normal temperatures should give no diffi- However, for all inspections performed before 1968 the
culties of stress corrosion cracking, and, consequently, very following simple technique was used, giving relatively low
little attention was given to the problem before 1964. At sensitivity: the surface under test was cleaned by a wire
that time, however, an almost new 230 m3 cylinderical tank brush and a thin coat of white paint applied. The magentiz-
started to leak through a 50 mm long crack, running along ing was done by sending 400-500 Amps AC directly through
a circumferential seam in the middle of the shell. The seam the material from contact electrodes spaced 150-200 mm
had been locally repaired over a length of 500 mm and apart. Black test ink was applied and the surface inspected
the crack was situated at one end of this repair. The leak under white light. A tangential field strength of 15-20
was observed 3 months after the repair was made and 3 Oersted was achieved by this system.
weeks after the tank had been filled up with ammonia.
This technique was improved in 1968 when the testing
A number of independent investigations were carried out of a spherical tank with numerous very fine cracks had
in order to find the possible cause of this and other similar shown the need of higher sensitivity.
cracks on the tank, and all results pointed to stress corro-
sion cracking. By diminishing the distance between the electrodes to
100 mm, a field strength of about 35 Oersted was obtained.
The tank was eventually repaired by renewing all welds At the same time the surface was cleaned more carefully
completely and stress relieving the complete tank before by machine brushing or by grinding, and the use of ultrav-
putting it back into operation. iolet light and fluorescent magnetic powder was taken into
use.
As a consequence of this incident it was decided to carry
out crack detection checks at regular intervals on all bigger A considerable amount of testing was done using this
storage tanks for liquid ammonia. The inspection should technique, but although many cracks were detected and
mainly cover the inside surface of the welded seams of the ground out, the most recent experience has shown the sensi-
tanks. tivity to be still insufficient. This became obvious when
a repeated inspection of two tanks, where cracks had been
During the period since 1964 an extensive inspection removed and later liquid ammonia had been stored over
program by means of the magnetic particle method and a period of 4 months, showed new cracks at the bottom
also to some extent by ultrasound has been carried out. of old grinding marks. A few test pieces containing such
The ultrasound investigation has particularly been used to cracks were cut out and a microscopic examination showed
look for defects and cracks which do not propagate to the all cracks to be following the deformation of the surface
interior surface of the tanks. layers, caused by the grinding operation carried out 4
months earlier. Apparently the grinding must have been
The tanks included in this program comprise a total of stopped before the cracks were completely removed. As this
8 spherical tanks out of which 4 are 5,250 m3 spheres, a was controlled by magnetic particle testting, the testing
large number of cylindrical tanks of different sizes, a few method must have been unable to indicate the cracks at
selected transport tanks of small dimensions, and 2 atmos- that instant. However, the influence of ammonia over a
pheric storage tanks, one of 25,000 m3 size, and the other period of 4 months must have widened the cracks so much
of 38,000m3 size. that magnetic detection again became possible. This theory

103
was supported by an experiment carried out on a few cracks number of cracks was discovered but it was ascertained
on one tank. The cracks were ground out to the extent by microscopic examination of test specimens that most of
that all magnetic indications disappeared. Following this, these had resulted from cracks which had not been com-
the surface was etched until the deformed surface layers pletely removed during the earlier repair.
were dissolved and finally a repeated magnetic testing was
made. Hereby the crack indication appeared again. In January/February 1970 the tank was investigated
100% by ultrasound. A total of 120 cracks and other defects
This led to the use of an even more sensitive method were disclosed, all positioned in the weld metal. About 80%
in which the tangential field strength was doubled to reach of the cracks were positioned at the crossing of weld seams
about 75 Oersted. This was done by use of 600-700 Amps and some of the cracks were quite deep.
through contact points not more than 70 mm apart. Only
fluorescent ink was used and all grinding out of cracks was Since it is not definitely known to which extent these
done by a sharp and fast running grinding stone only cracks may have originated before adding water to the am-
slightly pressed against the surface. This method has until monia it is still too early to draw any conclusion as to
now given satisfactory results and it has been checked on the effect of water addition.
cut out pieces on several occasions. A microscopic examina-
tion of such sample pieces has not disclosed any more
cracks than those found by the magnetoflux method. A fur-
ther improvement of the sensitivity of the magnetoflux is Appearance of stress corrosion cracks
possible and has been tested in the laboratory. Such further
increase in sensitivity has proven not to be desirable since The stress corrosion cracks appear in the inner surface
it gave evidence of apparent cracks on a sample which by of the welds and run in all directions, often as a branched
microscopic examination showed not to contain any cracks net of cracks. They do not penetrate into the adjoining
but only small structural differences in the weld metal with- plate. The depth of the stress corrosion cracks do not nor-
out any consequences for the strength of the weld. mally penetrate more than 5 mm measured from the inner
surface of the base metal except in cases where they may
connect to weld defects.

Case history Thermal stress relief is considered a most effective means

against stress corrosion.
5,250 m3 spherical tank manufactured 1964/65. The 38,000 m3 atmospheric storage tank was investigated
for stress corrosion cracks in 1969 after the tank had been
Material Si, Mn, Al-fine grain steel with a yield strength in service for about 5 months. The inspection concerned
of 33 Kp/mm2 and tensile strength 50-58 Kp/mm2 (1 a large number of welds, particularly in the lower part of
Kp/mm 2 = 1400 psi). the tank which is exposed to the largest tensions and where
the heaviest plate is found. No cracks were found during
Working pressure: 7 Kp/cm2
this inspection. Since plate dimensions and weld electrodes
Tank dimensions: 21.6 m°/22-28 mm correspond quite closely to those used for the spherical
tanks one would assume that a certain possibility should
exist that cracks might develop slowly. However, stress cor-
rosion cracking like any other chemical processes depends
When the tank was manufactured in 1964/65 it was X- strongly on temperature and at a temperature of -33°C
rayed and all cracks and other defects in the welds were the process is likely to go on so slowly that no stress corro-
repaired. Supplementary ultrasound investigations were car- sion cracking is to be expected even after years use of the
ried out of the complete lower horizontal seam, and of atmospheric tanks.
2,000 mm seam on each of the two middle horizontal
seams. No defects were found during this inspection. It is also worth mentioning that all of the thin walled
small transport tanks investigated, proved to be free of stress
.The first magnetoflux investigation was carried out in corrosion cracking. This may indicate the importance of
1968 and comprised all weld seams. Prior to the investiga- residual stresses in heavy plate tanks.
tion the weld seams were ground flush with base material;
then controlled for cracks. The investigation disclosed quite
a large number of fairly superficial cracks of which few Acknowledgement
penetrated more than 5 mm but there were also a few
cracks which penetrated up to 18 mm below the surface All information in this paper was made avilable by the
of the base material. Danish company A/S Ammonia through their representative
Mr. Preben Terkildsen. Most of the information is taken
The investigation was repeated after the tank had been from a number of reports prepared by the Danish Welding
in service for 4 months during which period the ammonia Institute for A/S Ammonia. The Danish Welding Institute
had been added 0.3% of water as a protection against fur- is an institution under the auspices of the Danish Academy
ther stress corrosion. At the new inspection a further large of Technical Sciences.