EMPTY SULPHURIC ACID TANK EXPLODES DURING HOT WORK AT TOP OF TANK

An empty 100-m³ tank sulphuric acid (H2SO4) exploded at 9:15 am at a chemical site. The tank was projected and fell nearby, toppling the scaffolding set up for reservoir maintenance and causing 3 workers (2 subcontracted personnel) to fall. Two of the three were seriously hurt: the plant employee sustained cranial trauma after falling some ten metres; and a subcontractor was pinned between the tank and the scaffolding, injured to the face by the grinder he had been using. The internal emergency plan was activated, bringing both internal and external responders to the site. The injured were taken to hospital. 

The tank had been temporarily plugged on 18th July using a sealant box after discovering a leak on 17th July. Scheduled for repair at the beginning of August following its drainage, the tank was rinsed with water throughout the weekend of 2nd August, then a subcontractor installed the scaffolding for the specific intervention: reservoir access, process insulation plating, etc. A plant employee, accompanied by 2 subcontractors, then climbed up the tank in order to open the top manway when the explosion occurred.

The accident appears to have followed a hydrogen (H2) accumulation at the top of the tank due to ignition of the inflammable mixture formed with air when cutting corroded bolts from the manway with a grinder. The tank was torn over half the circumference of the shell/bottom junction, and its anchorages were stripped. 
Insufficient rinsing of the tank (just a single rinsing cycle was performed) combined with the presence of a low concentration of sulphuric acid caused an acid attack of the metal, leading to the formation and accumulation of hydrogen at the top of the tank (dome-shaped tank roof). The explosion occurred by means of igniting the flammable mix created with air at the time of splitting the corroded bolts on the dome manhole with a grinder. The metallurgical assessment performed on this tank indicated the presence of extensive internal corrosion over the lower part of the structure. This observation confirmed the sudden onset of corrosion at the tank sidewall due to diluted acid, thus generating a source of hydrogen production.

Source:Aria database

ARE YOU OPERATING WITHIN SAFE OPERATING WINDOW?

When observing yellow smoke emanating from the chimney of a drying unit located downstream of the fertiliser plant granulator, an employee notified a control room technician; the workshop extraction fan was turned off at 10:30 am in order to limit discharges of both nitrous and chlorine gases, which were beginning to fill the workshop. The internal emergency plan was activated at 10:58 am, and fire-fighters arrived on the scene at 11:10. The dryer was started and then flooded; the incident was brought under control at 12:34 pm. The device was drained, with all fire extinction water collected in a retention basin; recovered sludge was recycled over the following week. 

Thermal decomposition had occurred inside a dryer, of the rotating tube variety, fed with hot air by a 7 MW/hr natural gas generator containing 20 tonnes of aggregates, over which an ammonium phosphate slurry had been sprayed. This ammonium phosphate supply was obtained by means of a chemical reaction between phosphoric acid (H3PO4) and ammonia (NH3). The site operator considered the possibility of accidental overheating, since the NPK 11-11-32 fertiliser had not been prone to self-sustaining decomposition. The unit had been shut down for the maintenance of a chain conveyor. The pertinent operating instructions would have been followed for this mission, given that the granulation loop contained dry matter, with the burner operating at the minimum setting (35%) and the drying drum no longer rotating.Subsequent to an excessive production temperature (> 300°C), the “dryer input” temperature, which was also abnormally high, surpassed the temperature at the onset of dry fertiliser decomposition (i.e. > 170°C). This thermal disequilibrium was caused by use of an H3PO4 at 38% concentration, which was more diluted than the normal level (53%). With a slurry containing a large quantity of water to be evaporated, gas temperature at the reactor output (set at 110°C) decreased, while the drying air temperature was automatically increased as a compensation to 300°C, by exceeding the typical threshold of 240°C. No alarm was triggered, with the 300°C value remaining below the 370°C threshold recorded on the “hot air intake” temperature probes. Afterwards, the dryer required more time to cool.The acidic dilution stemmed from an incident that had occurred 10 days prior, involving the unit’s three H3PO4 tanks, two containing a 53% acid and the other a diluted corroded acid (< 30%). With its shell leaking at a height of 1.5 m above the bottom, tank contents had been transferred into the other two tanks, thus diluting the acid used for manufacturing purposes.

The operator modified production standards, by introducing a "hot air intake temperature" alarm threshold adapted to each production run (260°C for ammonium nitrate fertilisers), along with the relevant maintenance shutdown procedure by indicating temperature controls and thresholds correlated with the steps required for installation shutdown, plus an internal emergency plan reminder to avoid stopping the fan in the event of toxic gas emissions.

Source: Aria database

ARE YOU INSPECTING ALL VULNERABLE COMPONENTS OF PIPING AS PER STANDARDS?

At 11:15 pm, high-pressure steam pipe elbow at 120 bar and 520°C connected to catalytic reforming equipment burst. The elbow and pipe , a 40-kg block of steel, was projected longitudinally. Inside the shop area, a walkway was ripped off its supports damaging an access ladder. It then flew over an ammonium nitrate conveyor belt only to land 230 m away in a parking zone for tanker cars full of ammonia awaiting shipment, which on that day happened to be empty. The steam tore apart the asbestos cement cladding on the wall located 20 m from the original rupture and escaped into the atmosphere with a load noise. 

The two employees present in the unit at the time shut down production operations and cooled the steam reformer with nitrogen. 

A metallurgical assessment of the elbow indicated that slow creep had initiated on the outer skin, which combined with flow into the material layer was the cause of pipe rupture. This hypothesis relied on the detection of oxidised, yet non-deformed, pipe openings in the presence of intergranular microcracks on both the pipe and its elbow. The origin of this creep was explained by the metallurgical composition of the elbow, i.e. ordinary carbon steel containing no alloys and not adapted to temperatures above 425°C. The pipeline was made of a slightly-alloyed P22 type steel, which was more resistant to creep and compliant with the original specifications defined 32 years prior for both material elements. 

Inspections carried out at the time on the elbow did not detect any noncompliance of the steel, given that non-destructive technology had not yet come of age. This creep might have been accelerated as a result of heat treatment performed at 700°C during equipment installation, once the assembly had been welded. Periodic inspections dedicated to pressurised equipment on the damaged pipe were only recorded into the log 25 years after service start-up, at the time of applying for facility recertification. The initial recertification was rendered official without any underlying structural documentation (misplaced), and subsequent inspections never focused on the section of pipe that would burst.

Source: Aria database

CONFUSION DURING A MAJOR LOPC CAUSES WRONG ACTION TO BE TAKEN

In a petrochemical complex, an ethylene (a highly flammable gas) compressor was in stable operation when a sudden drop in pressure occurred at 5:33 p.m., accompanied by a loud noise. The gas detectors in the zone became saturated, and a 2nd level alarm was triggered in the unit’s control room, as well as in the control room controlling the nearby compressor and at the safety station. The compressor began to vibrate. Its motor stopped but not its ethylene supply.

Confusion over which compressor was leaking
Not knowing which compressor was causing the leak, the shift crew of compressor No. 1 contacted the crew in charge of compressor No. 2. The latter crew persuaded them it was compressor No. 2 that was to blame. Crew No. 1 left their control room to help them. The alarms and parameters indicating the malfunction of compressor No. 1 (pressure drop) were not taken into account since the control room had been deserted.

The two shift crews, joined by the internal fire brigades, approached the area of the accident but were unable to enter due to the deafening noise. They encountered a flammable cloud of ethylene measuring 4 m high x 100 m, with visible droplets. The firefighters protected the nearby units with water curtains. Around 5:45 p.m., the shift leader 1 consulted with the shift leader in charge of the neighbouring unit. A leak on the ethylene supply network was suspected. Two operators, equipped with hearing protection, moved through the cloud, protected by a water curtain, to reach the network’s manual shut-off valves. They were able to close the valves manually, ending the leak. The cloud rapidly dispersed. The operators returned to the control room and closed the local supply valve of compressor 1. 8 t of ethylene (354 kg of flammable mass) was released in 21 minutes.

ROOT CAUSE: Valve ejection due to incorrect tightening and a non-compliant seal
The hatch and the valve porthole of the second stage of compressor 1 were found 6 m away. One stud from the hatch was severed (sudden brittle-type rupture), while the other five studs remained in place, three of which were missing their nuts and exhibited torn threads. These studs are compliant but had never been replaced since the compressor was commissioned 16 years ago. An expert assessment showed that the valve’s copper seal was not annealed at the time of its installation, contrary to procedure and the other seals on the equipment. It was therefore not as flexible and less able to absorb stresses. This defect, combined with a bolt tightening error on the verge of plastic deformation, led to fluttering in the stack and its rupture.

If the emergency stop had been activated, the leak could have been stopped more quickly as it shuts down both the motor and the ethylene supply. The operators believed that the motor shut-down because of vibration was sufficient. After the accident, the compressor was equipped with an emergency stop triggered by gas detection. However, the manual emergency stop remains in operation should this detection system fail.

Source: Aria database

INADVERTENT CHEMICAL ADDITION DURING CLEANING CAUSES EXPLOSION

 An explosion and fire occurred at night in a workshop set up to synthesise toluene diamine (TDA), by means of hydrogenating dinitrotoluene (DNT) in the presence of Raney nickel, during a scheduled maintenance downtime. In-house fire-fighters brought the fire under control within 35 min; in the meantime, four employees required hospitalisation. One of them, who was handling the valves to wash the hydrogenation reactors with isopropanol, sustained burns over 40%-50% of his body and died 15 days later. The workshop was completely destroyed. The reactor burst; the bunker housing the workshop was deformed due to the combined action of the blast wave and sprayed fragments; the reinforced concrete wall was ripped open, with rebar twisted; and the control room was heavily damaged. Glass panes were broken over a 50- to 100-m radius, while the distillation unit juxtaposing the bunker was damaged and allowed isopropanol and TDA to escape, adding fuel to the fire. Buildings belonging to the neighbouring industrial facility located 150 m away suffered deformations to their lightweight structures. 

According to the investigation conducted, this explosion resulted from injecting pure DNT into one of the reactors washed by the circuit used at the time of production startup. Two valves connected in series equipping this DNT feed line were found partially opened (at 10°) after the accident, most likely allowing 500 to 700 kg/h of product to flow into the reactor. The heat release upon hydrogenation of a small quantity of DNT probably triggered the sudden decomposition of the remaining DNT, while abruptly reheating the reaction medium. Corrective measures were adopted to prevent routing pure dinitrotoluene into the reactor. These were elimination of the DNT intake line on the injection tank, addition of two automatic on-off valves on the mixing tank’s DNT feed line, closure of the link (by an automatic on-off valve) between the mixing tank and the injection tank during the reactor washing