Flixborough Explosion

Flixborough Explosion

1st June 1974

28 Workers killed when an explosion occurred in a

plant processing Cyclohexane

Thanks to
Ann-Marie McSweeney &John Barrett
Department of Process Engineering, UCC
 Flixborough Explosion
Accident Overview
A temporary pipe was fitted between two sequential
reactors in a plant that was oxidising cyclohexane at
elevated temperature and pressure.
The pipe was not designed properly and the mechanical
loads acting on the pipe were not correctly identified.
In particular the pipe was subject to large bending loads for
which it had not been designed.
The pipe broke open at a thermal expansion bellows fitting
in the line.
Large amounts of liquid cyclohexane escaped through the
ruptured pipe and vapourised.
The vapour cloud found an ignition source and a fireball
ensued.
 Flixborough Explosion

Flixborough Explosion
A very good description of the incident is given in the book
MAJOR CHEMICAL HAZARDS
AUTHOR: V.C. MARSHALL
(In the UCC Library under Classification 660.28)

Also the Report of the Court of Enquiry.

The course notes for PE 2002 should also be consulted

especially the material dealing with bending of beams and
loading of pressure vessels.
 Flixborough Explosion
View of the Scene after the Incident
 Flixborough Explosion
View of the Scene after the Incident
 Flixborough Explosion
View of the Scene after the Incident
 Flixborough Explosion
PRODUCT DESCRIPTION
Raw material was cyclohexane (basically the alkane hexane with its
ends joined up!)

Formula C6H12 Molecular Weight M = 84

Boiling Point at Patm = 81 C

i.e. Cyclohexane is a volatile liquid with a low boiling point at ambient

conditions (something like petrol!)

Liquid Density 780 kg/m3 Vapour Density (at Patm) 2.4 kg/m3

Hence the liquid is lighter than water while the vapour is heavier than
air (in common with many hydrocarbons).
 Flixborough Explosion
PRODUCT DESCRIPTION
Cyclohexane has the following thermodynamic properties

Specific Heat Capacity Cp = 1.93 kJ/kgK

Ratio of Specific Heats = Cp/Cv = 1.087
Latent heat of evaporation = 360 kJ/kg
Flammability (in air) of the gas 5.3 % to 8.3 %
 Flixborough Explosion
PROCESS DESCRIPTION & REACTOR
CONFIGURATION

The Flixborough petrochemical plant was involved in the production of

cyclohexanone, a precursor for the manufacture of Nylon. The raw
material cyclohexane was oxidised to cyclohexanone by injecting air in
the presence of a catalyst.

The process of oxidation is slow and it was decided to use six stirred
reactors in series with the product from the first overflowing into the
second and so on. To do this the reactors were mounted on a platform
arranged in a series of steps each 0.355 m higher than the one following.

Good reaction kinetics dictated that the cyclohexane in the reactors be

maintained at the elevated temperature of 155 C. This temperature is
above its boiling point at atmospheric pressure so to hold it a liquid
state, the reactors had to be operated at 9 bar pressure.
 Flixborough Explosion
Oxidation of Cyclohexane

+ O 2

C y c lo h e x a n e C y c lo h e x a n o n e
 Flixborough Explosion
REACTOR CONFIGURATION
 Flixborough Explosion
REACTOR CONFIGURATION
This schematic view indicates the basis of the incident.
 Flixborough Explosion
PROCESS DESIGN FLAW
The cyclohexane in the reactors was in a liquid state at a temperature 74
C above its atmospheric pressure boiling point. Hence any loss of
containment would produce large scale flashing and escape of flammable
vapour!

In other words if any part of the reactor wall or associated piping broke,
the pressure would suddenly fall from 9 bar to 1 bar (atmospheric
pressure) and a huge amount of cyclohexane vapour would be generated.

Calculate of the proportion of liquid cyclohexane that would vapourise

(flash off).

Considering an adiabatic energy balance:

energy consumed in evaporating off vapour is provided by cooling of
the liquid fraction from 155 C down to 81 C.
 Flixborough Explosion
PROCESS DESIGN FLAW
Fraction Evaporated

About 40 % of the cyclohexane will vapourize. Given the inventory of

cyclohexane in the reactor train was about 100 tonnes, thus in the event of
an accident, 40 tonnes of vapour would be released.

Being heavier than air, cyclohexane if released would form a cloud in the
shape of an up-turned bowl. At the centre of the cloud, the vapour would be
close to pure cyclohexane but at the fringes, where it is mixed with air, the
concentration could lie in the flammable range.
 Flixborough Explosion
CONTAINMENT DESCRIPTION
An unwanted by-product of the oxidation reaction were some very
corrosive acids, which could only be contained by stainless steel.
However stainless can be 10 times the price of mild steel so the solution
was to make the reactors out of 12.5 mm thick mild steel with the
reactor insides lined with 3 mm thick stainless steel

The reactors were vertical cylindrical vessels with a diameter of

approximately 3 m and height of 6 m. The outsides were lagged with
thermal insulation protected by aluminium cladding.

Each reactor was adjoined to the adjacent reactors by a short stub pipe
of 0.7 m diameter. Due to the close proximity of the reactors to each
other, the length of this stub pipe was only about 1.5 m

There were a number of pressure relief valves in the system set to lift at
11 bar.
 Flixborough Explosion

INCIDENT DESCRIPTION
Sometime in March 1974, cooling water was sprayed on the outside of
Reactor 5 to quench a minor leak from a valve. However the water was
contaminated with chemicals which corroded the mild steel casing of
the reactor. The fact that the steel shell was under a tensile hoop stress
due to the contained pressure would have accelerated the damage (a
phenomenon known as stress corrosion).

This corrosion had the result that more of the mechanical load was
transferred to the stainless steel liner which was then overstressed and it
in turn cracked. Cyclohexane vapour began to leak from the reactor.

A first lesson of this would be that the system could leak as a result of
external corrosion (presumably not considered due to the lagging).
 Flixborough Explosion

INCIDENT DESCRIPTION
This reactor had to be shutdown and removed from service for repair.
To keep the process running, it was decided to fabricate a temporary by-
pass pipe to join Reactor Number 4 to Reactor Number 6.

POOR MECHANICAL DESIGN OF THE BY-PASS PIPE WAS THE

REASON FOR THE DISASTER
 Flixborough Explosion
BY-PASS PIPE GEOMETRY
The pipe in question (i.e. the by-pass) formed a connection between two
adjacent reactors over 6 metres apart. The reactor nozzles were
vertically off-set for process flow considerations so the pipe had to have
a dog-leg bend in it. The end of the pipe connecting to Reactor Number
4 was at a higher elevation (0.35 m) than the end connecting to Reactor
Number 6.

The by-pass pipe itself had a bore of 0.5 m though the stub pipes
emanating from both reactors were of a larger diameter of 0.7 m. There
were two bellows in the stub pipes to permit axial expansion or
contraction of the pipework.

The by-pass pipe was fabricated from stainless steel.

 Flixborough Explosion
By-Pass Pipe Geometry
 Flixborough Explosion

By-Pass Pipe Geometry

 Flixborough Explosion
Mechanical Properties
The mechanical properties of the by-pass pipe material would be
expected to be

Youngs Modulus E = 200 GPa

Coefficient of thermal expansion 18 x10 6 / C

Tensile Strength ts 600 MN / m 2

Steel Density s 7800 kg / m 3

 Flixborough Explosion
Membrane Stress in Pipe Wall due to Fluid Pressure
Because the pipe between the two reactors contains a pressurized fluid,
stresses will be developed in its walls.
PD
The hoop (circumferential) stress will be:
2t

Fluid Pressure P = 9 bar

Pipe Diameter D = 0.5 m
Pipe wall thickness t = 0.006 mm (In fact this is a guess; 6
mm is inch)

9 x 105. x 0 5 37 5 MN / m 2

2. x 0 006
This is certainly an acceptable stress level.
 Flixborough Explosion
Membrane Stress in Pipe Wall due to Fluid Pressure
Also have longitudinal stress in the pipe wall

PD
19 MN / m 2
4t

These were the only stress calculations carried out!

There was (and is) a piping design code which requires more rigorous
calculations and tests to be carried out on new piping. However (unlike
vessels) compliance with the code is not legally binding.
 Flixborough Explosion
BY-PASS PIPE BELLOWS REQUIREMENT
There were two austenitic stainless steel bellows at each end of the by-
pass where it joined the stub pipes (nozzles) from the reactors. This was
the fundamental fault with the arrangement yet they were essential. The
need to have a bellows in the pipe can be reviewed.

At times of plant shutdown, the pipe will be at ambient temperature

Tamb; say 15 C.

During plant operation, pipe temperature will rise to the operating

temperature of 155 C.
 Flixborough Explosion
BELLOWS REQUIREMENT
For a straight pipe with no bellows and assuming the reactor vessels at
either end act as perfectly rigid restraints, the induced thermal stress, T
assuming there is no accommodation of thermal expansion will be

T E. T

200 x 10 9. x 18 x 10 6. x (155 15)

504 MN / m 2

This is a compressive stress and is of such a magnitude as to certainly

cause damage to the pipe.
 Flixborough Explosion
BELLOWS REQUIREMENT
To avoid any thermal stress, the bellows must allow free axial thermal
expansion of an amount

T . T . L

18 x 10 6 x 140 x 6 5

16 4 mm

Note: A bellows is a flexible pipe section that allows large axial deflection
without generating high axial loads. Because of their construction a bellows
cannot tolerate any significant non-axial loads.

In the original system configuration where adjacent reactors were joined by

short straight lengths of piping, all the main pipe loads were axial. This was not
the case for the dog-leg by-pass pipe, a fact not taken into account at the time.
 Flixborough Explosion
Bellows Joint Illustration
The picture shows a pipe section with a bellows at either end (i.e. the
hoops)
 Flixborough Explosion
BY-PASS PIPE LOAD ANALYSIS
In addition to the membrane stresses in the wall of the pipe due to fluid
internal pressure, there are two additional mechanical loads.

Weight Loading
The normal span of the pipe running between adjacent reactors was
something over 1 m. However between Reactors 4 and 6 a span of 6.5
m was present. Thus bending due to weight loading (sagging) may be
significant.

The weight of pipe wall and of product inside the pipe can be found by
calculating the volume of each component and multiplying it by the
respective density:
 Flixborough Explosion
BY-PASS PIPE WEIGHT LOADING ANALYSIS

W 4
Di2 . L p 4 Do2 Di2 . L s g
Pipe inside diameter Di = 0.5 m
Pipe outside diameter Do = 0.5 + 2 x 0.006 =0.512m
Pipe length L = 6.5 m
Steel density s = 7800 kg/m3
Cyclohexane density p = 780 kg/m3

W 1021 484 . 9 81

W 14764 N
 Flixborough Explosion
Bending Moment due to Non-Collinear Fluid Pressure
Forces
The dog-leg bend in the pipe means that a moment is developed due to
the equal, opposite, parallel and non-collinear fluid pressure forces that
the pipe is subject to at either end.

The Bending Moment due to the pressure forces, Mp is

M p Fp . e
e pipe offset (or eccentricity) = 0.4 m

The pressure forces, Fp are:

F p p. A p Di2
4
F p 9 x 105. . 0 7 2 3 46 x 105 N
4
 Flixborough Explosion
Bending Moment In By-Pass Pipe
Note the diameter of 0.7 m is taken because fluid pressure is developed
at the original stub pipes.

M p 3 46 x 105. 0 4

1 4 x 105 Nm

This was not included in stress calculations carried out on the by-pass
and is of critical importance! This Bending Moment by far the larger of
the two loads.
 Flixborough Explosion
Beam Analysis of By-Pass Pipe
Having identified the loads, the response of the by-pass pipe to them
can be studied.

The actual by-pass pipe support was quite complicated with the
arrangement being supported by scaffolding poles. Also because of the
two dog-leg (mitre) joints, the cross section is not constant.

Treat the pipe/reactor flange and adjacent bellows as a simple support

Provides resistance to deflection.
Provides no resistance to slope (because of the bellows).

Treat the by-pass pipe as a straight beam.

 Flixborough Explosion
Beam Analysis of By-Pass Pipe
So the structure corresponds to a simply supported beam subject to a
uniformly distributed weight load and a moment at the centre of the
span.
The magnitude of the uniformly distributed load, q is

W 14764
q 2271 N / m
L 65

The magnitude of the applied bending moment, Mp is

5
M p 1 4 x 10 Nm
Mp q
 Flixborough Explosion
Beam Analysis of By-Pass Pipe
A Shear Force and Bending Moment diagram for the beam can be drawn. Firstly must
calculate the reaction forces at either end of the beam, R A and RB respectively by
considering equilibrium of the structure.

Equilibrium of Vertical Forces

R A RB q. L

Equilibrium of Moments about end A M A 0 RB . L M P 1 2 qL2 0

M P qL qL M P
kN 28 9

RB RA
L 2 2 L

The reaction force is largest at support B i.e. at the bellows attached to reactor number 6.
At this end, the load due to supported weight and the bending moment sum together.
1 4 x 105 2271. 6 5
RB
65 2 RB 21 5kN 7 4kN

R A 7.4 21.5 14.1 kN

Note that the reaction force at support A is
 Flixborough Explosion
Shear Force & Bending Moment Diagram for Pipe
M P

R A q

R B

x
0

1 0

2 0

3 0

8 0

7 0

6 0

5 0

4 0

3 0

2 0

1 0
x
0

-1 0

-2 0

-3 0

-4 0

-5 0

-6 0

-7 0
 Flixborough Explosion
PIPE RUPTURE
The above calculations indicate that the bellows at the lower end of the
pipe (adjacent to Reactor Number 6) was exposed to a non-axial (in fact
a perpendicular or transverse) force of almost 30 kN.

This would have been sufficient to rupture the bellows and allow large
quantities of cyclohexane to leak out.

This was agreed as being the probable cause of failure.

What is known is that on the date in question, this by-pass pipe running
between reactors Numbers 4 and 6, ruptured and released in the region
of 40 tonnes of cyclohexane vapour. The cloud subsequently ignited
probably due to contact with open flames in an adjacent Hydrogen plant
and exploded.
 Flixborough Explosion
FIRE BALL (BLEVE) CALCULATIONS
Model the instantaneous combustion of the escaped vapour. Duration of
burning of fire ball is

td Duration of fire ball s

M Mass of fuel in fire ball kg

The radiative power of the fire can be calculated from

QR Radiative power W
HC Calorific Value J/kg
 4
L 3 t

Flixborough Explosion
FIRE BALL (BLEVE) CALCULATIONS
A point source model of the fire gives the radiative heat flux as

Radiative flux W/m2

QR Radiative power of flame W
r Distance from source m

In turn the thermal radiation dosage can be calculated as

L Thermal radiation dosage (kW/m2)1.33s

Intensity of radiation (radiation flux) kW/m2
t Duration of exposure s
 4
L 3 t

Flixborough Explosion
FIRE BALL (BLEVE) CALCULATIONS
Note the duration of exposure is equal to the duration of the fire ball.

Damage to people exposed to the fire can be quantified with

Hence can estimate how close people must have been to the fire to have
been killed or injured.
 Flixborough Explosion
EXPLOSION AFTERMATH
Picture shows reactors 4 and 6 after the accident.
 Flixborough Explosion
EXPLOSION AFTERMATH
Picture shows the by-pass pipe after the accident
 Flixborough Explosion
CONCLUSIONS
28 workers were killed and 36 injured on the site.
53 people were injured off-site.
1821 houses were damaged.

Lessons
The accident occurred due to:

Poor Process Design

Lack of Understanding of Mechanical Loading of Process Equipment