A chemical plant for the production of fertilizer was built in 1969 and located over 7 miles from the town of Jonova which had a population of 40,000. The plant employed 5,000 people and was positioned inside a military zone with controlled access. The ammonia tank involved in the accident was a cryogenic storage vessel with a capacity of 10,000 tons (t) (inside diameter was 30m with a height of 20m), of Japanese design but built by the Soviets in 1978. The reservoir had a single wall insulated with perlite held in place by an external skin of steel. The gap between this skin and the reservoir wall was filled with nitrogen under low pressure. This tank, designed to resist an internal pressure of 1.5 psi, stored liquid ammonia (density: 0.68) under slight overpressure (1.2 psi) at a temperature between -26°F and -29°F.
Fed by a production factory (1400t/day) at 3/4 of a mile distance, the reservoir was close to 3 storage sites containing a total of 55,000t of fertilizer
The other equipment comprising the unit was as follows:
- Two piston compressors with a transfer capacity of 323 m3/h, one with an electric motor, the other with a diesel engine. These could be used when the two turbo-compressors used for the transfer of the ammonia from the production unit to the reservoir were halted.
- A flare-stack with a capacity of 500 kg/h,
- Two valves each with an evacuation flow rate of 4 200 m3/h,
- Two safety valves protecting the reservoir against depressions,
- A monolithic reinforced concrete protective wall 14.1m high and 400mm thick, also used as retaining tank and dimensioned to bear the hydrostatic pressure of the liquid.
- Alarm systems relating to the pressure and the liquid level in the reservoir.
- The accident
On March 20th, 1989, the pressure in the tank climbed abruptly between 11:00 a.m. and 11:15 and it burst at its base. Under the effect of the liquid ammonia escaping from the gaping breach, the tank broke free from its stand, pushed in the opposite direction. It destroyed the reinforced concrete protecting wall and ended up 45 yards from its foundation.
The 7,000t of ammonia contained in the tank spread over the ground forming a layer which was, in places 28″ deep.
With only a light wind (<2 m/s), it took 12 hours to evaporate.
According to the local authorities, part of the liquid ammonia was propelled, in the form of a jet, towards the phosphonitrate production buildings. A fire then spread to the fertilizer depots. Other sources claim that the flare-stack ignited the ammonia cloud, the flames were then propagated to all the buildings on the site. The collapse of a burning conveyor belt onto a store of fertilizer (11-11-11) would then have initiated the decomposition of the compost which continued for three days while releasing large quantities of nitrous oxide (NOx) to the atmosphere.
The emergency services arrived 30 minutes after the accident; the means deployed increased progressively: civil defense, the emergency commission of the Lithuanian Republic and the military region, civil defense teams from the USSR and the
Soviet ministry of fertilizers.
The official human casualty list cited seven deaths and 57 wounded (treatment lasting from 2 to 3 weeks) among the operational personnel of the plant and of the construction companies working close to the accident site. The municipal authorities alerted 25 minutes after the beginning of the accident, decided to evacuate the high-risk areas as soon as the concentration of ammonia in the air exceeded 10 mg/m3; 32,000 people were thus displaced.
A toxic cloud composed of ammonia vapor and products of the thermal decomposition of the fertilizer (nitrous oxides, ammonia) was at the origin of irritation observed as far away as 22 miles from the accident site. The contaminated zone
extended to 250 square miles.
The estimated height of the cloud at 3, 6 and 12 miles from the site appears to have reached respectively 328′, 1300′ and 2600′.
According to the report from the Soviet authorities, the concentration of ammonia along the track of the cloud and within a radius of 2 miles did NOT exceed 200 mg/m3
The levels found at a distance of 6-9 miles were between 20 and 40 mg/m3
These relatively low levels of ammonia could be explained, at least in part, by the ignition of the ammonia vapor in the plant.
The maximum distance over which the presence of ammonia could be detected in the air was 14 miles.
The concentration of nitrous oxides (NOx) which could have reached 25 mg/m3 on the site close to the storage zone for the phosphonitrates, did not exceed 2 mg/m3 in the wake of the cloud that had formed.
To reduce the impact of the cloud on the surrounding areas, water curtains were set up using fire hoses all along the track of the gas cloud. Furthermore, special measures for the protection of watercourses were taken to avoid the risk of pollution of the nearby NERIS river.
Pumping of the extinction water, the positioning of containment basins…
THE ORIGIN, CAUSES, AND CIRCUMSTANCES OF THE ACCIDENT
The day of the accident, one of the liquefying turbo-compressors used for the transfer of the ammonia from the production unit to the cryogenic reservoir was halted for long-term maintenance. Around 10 a.m., the second turbo-compressor was suspended for a repair job of short duration.
The operators then put into service the safety piston pump to draw the gaseous ammonia from the production unit, but difficulties in starting up the cooling circuit for the compressor delayed this operation. The pressure inside the storage vessel at that moment was at 1.0 psi. The ammonia coming from the production unit was then diverted to the flare stack. Despite these measures, according to the official report, nearly 14t of warm ammonia (+10C) were introduced into the lower part of the cryogenic reservoir (corresponding to 15 minutes of transfer).
According to the official information, these 14t of NH3 injected into the bottom of the reservoir formed a bubble in the lower part of the reservoir under the effect of the hydrostatic pressure. The warm ammonia may then have reached the surface (a
phenomenon comparable to roll-over) causing a sudden increase of the pressure in the reservoir, exceeding the released capacity of the valves and causing the rupture of the reservoir, more than an hour after the introduction of the warm ammonia.
Nonetheless, according to Anderson, this hypothesis seems unlikely, taking into account the pressure differential between the liquid ammonia at the base of the reservoir (relative ≈ 16 psi) and the ammonia at 10C that was injected
(relative 73-87 psi) ; the hydrostatic pressure was certainly insufficient to prevent the vaporisation of the warm ammonia at the at the bottom of the reservoir and its rise to the surface. The calculation of the mass of vaporized ammonia (2,52 t) and consequently of the volume of gas generated (2919 m3 @ -33°C) shows that the protection valves, capable of evacuating 2016 m3 of gas in 15 min, were not sufficiently dimensioned to protect the reservoir in this precise case. Internal pressure had thus probably increased, provoking the rupture of the retaining shell.
Inquiries conducted following the accident revealed that:
- the stronger resistance of the roofing of the reservoir as compared with the attachment of the inside walls of the tank to the base, as well as of the gusset plates, caused the rupture of the reservoir at its base. The floor of this remaining solidly attached to its foundations,
- the wave of liquid ammonia released broke the protecting wall, before spreading out over a wide area, thus aggravating the consequences of the accident,
- the resistance of the protection wall was not in conformity with the specifications drawn up in the design of the unit (real breadth inferior to the 400 mm specified) on account of modifications made during construction to reduce the materials and labour costs. During development, it appears that modifications were also made, for the same reasons to the foundations of the reservoir and its anchoring devices.
Finally, the official report speaks of “unfavorable operating conditions” in particular as regards the liquefying turbocompressors.
THE MEASURES TAKEN
To deal with the causes at the origin of the accident which occurred on the Jonova site, the report from the Soviet authorities mentions various technical measures relating to the design, construction, and operation of cryogenic reservoirs of this type:
- reinforcement of the protecting walls around the storage tanks, so that they can resist the dynamic pressure resulting from a maximum volume released in the event of shattering of the reservoir,
- continuous recording of the significant variables involved in cryogenic storage, with duplication and recording in the control room,
- automatic taking out of the circuit the ammonia feed at the base of the reservoir as soon as the temperature rises above -30°C and filling of the reservoir from the top,
- automatic diversion of the gas towards the flare stack when the pressure in the tank exceeds 800 mm column of water, the handling capacity of the flare stack having to be at least equal to 20 000 Nm3/h,
- storage capacity for ammonia limited to 80 % of the volume of the cylindrical part of the reservoir,
- installation of non-return valves and elimination of certain sections of piping to prevent the arrival of warm ammonia into the cryogenic reservoir,
- remote start-up of the re-capture pumps for ammonia spilled into the retention tank.
THE LESSONS LEARNED
This major accident shows that the ruin of a cryogenic reservoir is possible, as is the ignition of an unconfined cloud of ammonia (without explosion) although this phenomenon has rarely been reported in accident science.
The dramatic consequences of this accident result from the rupture of the reservoir at the level of its base, with the release of its entire contents. On the other hand, the rupture of a cryogenic ammonia reservoir in Geismar (USA) in 1984 at the level of the upper ring die did not lead to the release of liquid ammonia:
the only consequences observed were the presence of 150 to 400 ppm of NH3 downwind from the storage site during the 6 hours following the accident.
The adequate dimensioning of the resistance of the reservoir at the level of their base and their summit (the frangible nature of the reservoir) can thus allow for the limitation of the consequences of severe over-pressure. Roll-over phenomena may, furthermore, be avoided by feeding the reservoir from the top or equipping the tank with an internal re-circulation system. The accident which occurred at Rostock (Germany) in 2005 provides another example of the violence of the thermodynamic phenomena which can occur in cryogenic storage : the formation of two strata (one of ammonia at 20 % and the other of anhydrous ammonia) in a cryogenic reservoir during filling, on account of the presence of oil, caused the rupture of the tank during a sudden mixing of the two layers.
These two accidents illustrate the misleading phenomenon of inertia associated with large-scale storage and the low level of thermal exchange which can generate a too rapid mixture of ammonia in different phases.
To limit the risk of over-pressure in cryogenic reservoirs, it is essential to maintain a sufficiently low temperature (-33°C) during storage and also during filling even if small quantities are involved relative to the volumes stored (14 t à 10°C in the present case as against an amount stored of 7,000 t). Protective equipment such as such as bursting discs and valves, correctly dimensioned, must also play their role. Nonetheless at La Madeleine (59) in 1989, the rupture, belatedly detected, of a security disc equipping a cryogenic reservoir provoked the release to the atmosphere of 2.4 t of NH3 and therefore it is useful to examine the evacuation conditions for released materials preventively.
More generally, information of several kinds can be drawn from this accident:
- Maintenance periods are always risky situations which demand increased vigilance, as do exceptional circumstances or special operations. A prior examination of the risks, proportional to the stakes, should be conducted and compensatory measures should be formally taken.
- Even if it considered as passive, the retaining equipment might turn out to be ineffective, as against the dynamic effects of the flow of liquid (submersion) or when it is damaged by the impact of the wave or by domino effects (projectiles).
- Safety measures and equipment, such as emergency controls must be permanently accessible, even in degraded situations.
- The redundancy of the equipment (2 turbo-compressors available + 1 piston pressure pump in the present case) reduces the probability of a major accident but does not exclude it : multiplication of the various types of barrier (alarms, servo-systems, emergency operating procedures, fail-safe security systems…) while avoiding common sources of failure mainly when the stakes are high, or the installations are particularly dangerous, are the most effective way of reducing significant risks, without claiming to eradicate them. Bearing in mind the quantities stored in cryogenic reservoirs, the potential sources of danger in these installations are numerous; therefore the measures taken on the technical, organizational and human fronts to prevent accidents should be proportionate to these.