2. Introduction
• Largest peacetime explosion ever to occur in the UK
• Date: Saturday, 1 June 1974
• Location: Flixborough chemical plant owned by Nypro
(UK) Ltd
• Deaths of 28 workers on the site
• Widespread damage to property within a 6 mile radius
around the plant
3. • 1967: 20,000 TPA caprolactam plant built by DSM at the
Flixborough site using process involving the hydrogenation
of phenol
C6H5OH + 2 H2 → (CH2)5CO
• 1972: 70,000 TPA ; new process used
• New process based on oxidation of cyclohexane
• Posed much greater hazard than phenol process
4. Description of the cyclohexane
process
• Process operates by injecting compressed air into
liquid cyclohexane at a working pressure of about 9
bar and temperature of 155°C
• Cyclohexanone and cyclohexanol produced
• Conversion is low and it is necessary to recirculate
the cyclohexane continuously through a train of six
large SS-lined reactors
7. Events leading to the accident
2. Miners overtime ban of Nov, 1973
• Resulted in the government passing legislation to restrict
the use of electricity by industry to 3 days a week
• Was not possible to operate the process on this basis
• Was decided to utilize existing emergency power
generation on-site
8. • Major electricity user: 6 stirrers in the cyclohexane
reactors
• Primary purpose: disperse compressed air that was
injected into each reactor via a sparger
• Also ensured that droplets of water formed within
the reactor system were dispersed into the
cyclohexane
9. 1. The No. 5 reactor problem
• Jan, 1974:normal electricity supply resumed
• Was found that the drive mechanism for the stirrer
in the No. 4 reactor had been subject to severe
mechanical damage
• No reason was found for this. It was therefore
decided to continue to operate the plant with the
No. 4 reactor stirrer shutdown.
10. • Cyclohexane reactors were MS vessels fitted with an
inner SS lining to resist corrosion
• March,1974: Cyclohexane found leaking from 6 feet
long vertical crack in the MS shell of the No. 5 reactor
• Due to technical problems experienced earlier and the
effects of the 3-day week, the plant owners were keen
to make up lost production
• Therefore decided to remove No. 5 reactor for
inspection and continue operation with the remaining
five reactors
11. 1. Installation of 20” bypass pipe
• This pipe connected together the existing 28 inch
bellows on the outlet of reactor No. 4 and the inlet of
reactor No. 6
• Dog-leg shape of pipe
• Company did not have qualified mechanical engineer
on site to oversee design and construction
• No hydraulic pressure testing of pipe carried out,
except for a leakage test using compressed air.
13. 1. Resumption of production
• Plant restarted and operated normally, with occasional
stoppages, up until the afternoon of Saturday, 1 June 1974
• Previous day: plant had been shut down for minor repairs
• Early hours of 1 June: plant in process of being restarted
• Start-up involved charging system with liquid cyclohexane to
normal level and then recirculating this liquid through a heat
exchanger to raise the temperature.
14. • The pressure in the system was maintained with nitrogen
at about 4 bar until the heating process began to raise the
pressure due to evaporation of cyclohexane.
• The pressure was then allowed to rise to about 8 or 9 bar,
venting off nitrogen to relieve any excess pressure. The
temperature in the reactors by then was about 150°C.
15. • On 1 June this procedure was followed except it was noted
by the morning shift that by 06.00 hours the pressure had
reached 8.5 bar even though the temperature in the No. 1
reactor had only reached 110°C
• Was not realized at the time that this discrepancy might
have indicated the presence of water in the system
• The start-up continued until, at about 16.50 hours, a shift
chemist working in the laboratory close to the reactors
heard the sound of escaping gas and saw a haze typically
associated with a hydrocarbon vapour cloud.
16. The accident
• 16.53 hours on 1 June 1974: massive aerial explosion
occurred with a force later estimated to be about 15 to
45 tonnes of TNT equivalent
• Explosion heard up to 30 miles away and damage
sustained to property over a radius of about 6 miles
around the plant
• 28 plant workers killed with no survivors from the
control room
• All records and charts for the start-up destroyed
• Following the explosion, 20 inch bypass assembly was
found in a ruptured condition
17. The Public Inquiry
• Following the disaster, public inquiry
conducted under the chairmanship of Roger
Parker QC
• To establish the causes and circumstances of
the disaster
• To identify lessons to be learnt from the
disaster
18. Conclusions of the inquiry
• The immediate cause of the main explosion
was the rupture of the 20 inch bypass
assembly between the No. 4 and No. 6 reactor
• Two main theories to explain
19. The 20 inch pipe theory
• The 20 inch bypass assembly failed due to its
unsatisfactory design features
• However, the assembly had survived 2 months of
normal operation.
• A number of independent pressure tests were
commissioned to determine unusual conditions
20. • The normal working pressure = 8 bar
• practice during start-up to allow the pressure to
build up to about 9 bar.
• The safety valves for the system, were set to
discharge at a pressure of 11 bar
• At pressure above 11 bar, squirming motion
which distorted the bellows.
21. • Even when the assembly squirmed, no rupture
until pressure crossed 14.5 bar, a pressure not
achievable in reactors.
• Inquiry concluded that a rupture of the 20 inch
bypass due to pressure, temperature conditions
• Report conceded ambiguity in the hypothesis
• Simulation tests could not replicate failure at
similar conditions
22. The 8 inch hypothesis
• Alternative theory
• 50 inch split in an 8 inch line connected to
separator below bypass
• this failure led to a smaller explosion causing
failure of the main 20 inch bypass
• Zinc embrittlement had caused the split
• Small lagging fire at a leaking flange causing zinc
to drip onto the 8 inch pipe
• Brittle failure – Vapour release – Explosion – 20
inch failure
23. • Inquiry Report had devoted discussion of this
two-stage theory
• Finally dismissed as being too improbable
• No other theories considered by them to explain
failure of 20 inch bypass pipe
24. The water theory
• Another alternative theory
• Not considered by the Inquiry
• Much of scientific work after Inquiry closed
• Examined the effects of not operating the No. 4
reactor stirrer during the start-up at a time when
water may be present
• More probable explanation
25. • Cyclohexane and water are normally immiscible
• Azeotrope forms due to the limited solubility of
water in cyclohexane.
• This azeotrope has lower boiling point than
either water or cyclohexane
• Unstable interfacial layer may form
• Under certain conditions can boil and erupt
violently ejecting cyclohexane and superheated
water from the reactor.
26. • Normally impossible for water layer to form due to
dispersion of water by air distribution
• During start-up, the air to the reactors shut off
• If stirrers running during start-up, no water layer
• If stirrer stops, a layer of water could form, together
with the unstable azeotrope.
27. • As temperature of reactor increases, boiling
point of azeotrope is reached
• Possibility of a sudden violent eruption from the
reactor and ejection of slugs of liquid reactant
• Slugs exert high mechanical forces on the bypass
assembly, loosely supported by scaffolding
• Causes bypass assembly to fail without the high
static pressure in the reactors
28. Alternative event sequence
• Most credible explanation
• Explains failure of 20 inch bypass
• Also provides an explanation for the whole
sequence of events
29. • Unexplained failure of drive mechanism of No. 4
reactor
• Crack developing in the lining and shell of No.
5 reactor
• Failure of the 20 inch bypass assembly.
• Any/all failures caused by violent eruption of
reactor contents due to presence of water
• Committee failed to see common thread
30. • Drive mechanism failure for No. 4 reactor
unexplained
• Thought to be irrelevant
• Crack in the shell of the No. 5 reactor due to
stress corrosion
• Proposed by plant owners but not credible
31. • The failure of the bypass concluded by Inquiry
due to reactors being over-pressurized
• Implies human error, not verifiable
• Greatest failing of Inquiry was not taking
account of all the events 6 months prior to the
disaster
• Issue of non-operation of the reactor stirrers
ignored
32. Conclusions
• Human error analysis
Table gives causes against the different types of
error that occurred.
• Direct cause
Failure of the 20 inch bypass pipe led to huge
release of inflammable cyclohexane vapour
which ignited
• Root causes
A badly designed 20 inch bypass pipe installed
rather than finding reasons for the crack in the
No. 5 reactor
Why bypass failed ?
34. Safety considerations
• Learnings
• Low inventory especially of flashing fluids
• Before modifying process, carry out systematic
search for possible cause of problem
• Carry out HAZOP analysis
• Construct modifications to same standard as
original plant
• Use blast-resistant control rooms and buildings
