Wednesday, April 24, 2013

Flixborough Disaster - Explosion of a Cyclohexane Cloud

Flixborough Disaster

Explosion of a Cyclohexane Cloud

Summary data:
Date: June 1st, 1974
Place: Flixborough, United Kingdom
Type of accident: Explosion in a cyclohexane processing plant
Outcome: 28 deaths (workers), 89 injured  (36 on site, 53 off-site), 1821 damaged houses. Loss about $66 million which is equivalent to $200 million today .

Almost 40 years after the explosion at the Flixborough chemical plant, the mystery remains as to the true causes of this accident. 


View of the Flixborough chemical plant after the explosion
  
The story began in 1938 when a plant to produce fertilisers was built near the village of Flixborough. Then, it evolved in 1964 to produce caprolactam, a precursor for the manufacture of Nylon. The process used the hydrogenation of phenol. In 1972, a new unit was implemented to increase the production to 70000 tons per year. The technology was based on the oxidation of cyclohexane into cyclohexanone : this process was known to be more hazardous than the phenol process.

Nevertheless, the company Nypro, owner of the site, was experiencing serious economic problems. In reality, the unit only produced 47 000 tons of caprolactam per year. Last but not least, the government had been with a miners’ strike since 1973. They declared a state of emergency and established a three day working week to save electricity. It was not possible to operate the process on this basis. So emergency generators were used to run the essential equipment and other equipment, such as the six stirrers in the cyclohexane reactors, was stopped.

Overview of the accident: The largest industrial explosion ever in the UK

To understand the accident, the diagram below represents the unit concerned and its process.
In the six in-line stirred reactors, the cyclohexane is oxidized by injecting compressed air. The process operates at 8,8 bar and 155°C and produces cyclohexanone and cyclohexanol. But this reaction yield is low, therefore, the cyclohexane is recirculated to feed it again.
Configuration of the cyclohexane oxidation process

On Saturday 1 June 1974 at 4:53 pm, the plant and all the surrounding buildings within 600 meters were devastated. The deflagration was heard up to 50 kms away. This accident caused the death of 28 workers, with no survivors from the control room, and 89 people were injured: 36 on site, 53 off-site. The fires burned for several days. The only ‘positive’ point in this tragedy is the accident happened during the week-end. On a regular weekday, 550 employees worked there. 
By-pass pipe between reactors 4 and 6


The immediate cause of the main explosion was the rupture of the 20 inch by-pass assembly between reactors 4 and 6.  40 to 60 tons of cyclohexane escaped and were ignited 30 seconds after by the reforming tower forming a cloud two hundred meters in diameter and 100 meters in height. Before this explosion, on 27 March, a vertical crack in reactor 5 had been discovered leading to a cyclohexane leak. A serious problem was identified, so this reactor was removed and replaced by a simple by-pass between reactors 4 and 6.


 Causes: No witnesses … Several theories

Despite the investigation lead by the Secretary of State for Employment, no unanimous conclusion could be drawn. However several theories were born to explain the rupture of the 20 inch by-pass assembly between reactors 4 and 6. 

The 20 inch pipe theory
According to the inquiry, the 20 inch pipe broke because of temperature and pressure conditions higher than usual, even if they were within the range of operating conditions. Simulation tests could not replicate failure at similar conditions, so an ambiguity remained. The real cause remained the assembly pipe between reactors 4 and 6 since no study and no test were launched before its installation. The problem on reactor 5 should have been studied before running the unit again. In fact, the reactors had been sprayed with water to dilute the cyclohexane leaks to limit flammability, but the nitrates of the water dripped into the steel pipe and caused to corrosion.

The 8 inch hypothesis
An alternative theory was studied, based on the 8 inch pipe hypothesis. It stated that the first explosion could have been due to a cyclohexane leak on an 8 inch pipe causing failure of the main 20 inch by-pass. The problem on the 8 inch pipe could have been due to faulty seals or check valve, bad fitting of two bolts, contact between zinc and steel, leading to a double leak of cyclohexane and then the cracking of the 20 inch pipe.
This theory was abandoned because the succession of events was too improbable.

The water theory
The last theory was not considered by the inquiry since it was formulated by scientists who continued to  work on it after its closure. The rupture of the 20 inch pipe could be due to a sudden rise of pressure because of the water presents in reactor 4. An azeotrope could have formed between the water and the cyclohexane. Normally there is no water in the reactors but on the day of the accident, the stirrer (reactor 4) had a mechanical problem and was stopped:  water may have been present.
Cyclohexane and water are normally immiscible. The stirrers normally prevent the water from having a low solubility in the cyclohexane but, that day, two phases may have occurred with an unstable interfacial layer, the azeotrope. At start-up, the temperature increased, the boiling point of the azeotrope was reached and this may have led to a sudden pressure rise and the cyclohexane ejection. The bypass failed under the high pressure in the reactor.

Lessons from the accident

Facilities design: The place and the structure of the control room have to be carefully chosen. The best solution is to be as far away as possible from the more hazardous units and to minimize the quantity of dangerous products on the site.

Process modifications: For hazardous processes, every little change has to follow the same standards. Good industry practices require that modifications should not be carried out without having undertaken a safety, engineering and technical review. In this case, there was no professional engineer in the plant at the time of the accident. For example, they needed a 28 inch pipe to join reactors 4 and 6, but they only had a 20 inch pipe… they fitted it with strapping and a plate instead of shutting down the unit.

Maintenance planning (anticipation): It is essential to establish a planning to anticipate the problems. If a problem occurs temporary repairs are not the solution, particularly if it stays in place 2 months (as was the case). The startup of a unit after a problem must only take place after identifying the causes.

Human analysis: The employees who are responsible for the unit should know the unit, the risks and how to find the basic documents in case of a problem (technical documents, design specifications …). Experienced people may recognize the precursors of an accident.
Last but not least, the employees do not have to choose between safety and production: safety comes first.

    Sources:
  - Explosion catastrophique d’un nuage de cyclohexane, le 1er juin 1974, Aria - Ministère chargé de l'environnement, fiche mai 2008, available at http://www.aria.developpement-durable.gouv.fr/ressources/5611_flixborough_eb_fr1.pdf
   - Major Technological Risk, An assessment of Industrial Disasters – P. Lagadec, translated by H. Ostwald, available at http://www.patricklagadec.net/fr/pdf/Amoco_Cadiz_EN.pdf



5 comments:

  1. What is the 8 inch pipe connected to? It's a little confusing.

    It's a good summary with clear and well organized causes and lessons learnt.

    ReplyDelete
  2. Was there any security department in charge of validating the work replacement that was done. There is usually procedures to follow, was it the case here ?

    The different theories are clear and all seems plausible !

    ReplyDelete
  3. Recent years, scientists and researches make exploitation to ultrasonic crushing technology and produced a set of ultrasonic crusher. From here: www.toption-china.com/products/ultrasonic-cell-crusher-ordinary-type. This development can improve the crushing effect of mineral stone with smoother and faster vibration.

    ReplyDelete
  4. Glass cutters don't work by actually cutting surface, they make a very fine score that begins the break. Getting a good score is key to getting a very clean break. I think the basic cutter with a straight shaft and oil reservoir is a good option. gözlüğü

    ReplyDelete
  5. Movie leaks vs cyclohexane leaks

    ReplyDelete