Gas Explosion Generated Drag Loads in Offshore Installations
Gas Explosion Generated Drag Loads in Offshore Installations
Gas Explosion Generated Drag Loads in Offshore Installations
Abstract
There is little guidance in place for secondary blast effects for offshore installations. In this paper, the origin of drag loading
due to blast is explained, including the different nature of loading that could be expected for a range of explosion scenarios. The
drag forces on pipe work, pipe supports grated decks and equipment are discussed. Methods on how to calculate these forces are
described. 1998 Elsevier Science Ltd. All rights reserved
Keywords: Offshore installations; Drag loading; Temporary Refuse; Offshore safety case; Gas explosion
0950–4230 /98 /$19.00 1998 Elsevier Science Ltd. All rights reserved. Printed in Great Britain.
PII S 0 9 5 0 - 4 2 3 0 ( 9 7 ) 0 0 0 5 4 - 5
44 R.B. Corr, V.H.Y. Tam / Journal of Loss Prevention in the Process Industries 11 (1998) 43–48
end, the resultant flame would move towards the open also occurs but the mechanism for generating it is differ-
end. As the gas burns, it expands to about seven times ent. In this case, the over-pressure region tends to be
its initial volume. This excess volume is vented out of confined and propagates with the flame. As the driving
the box—thus the explosion produces a wind. If the force for the flow is over-pressure, the flow reverses after
venting is not sufficiently fast to remove all the excess the passage of the pressure wave.
volume generated by the combustion, then an over-
pressure will develop inside the box. 2.4. Drag forces
In the above example, the gas flow speed is highest
close to the open end (vent) and lowest at the closed end. This flow reversal also produces drag force direction
This distribution of gas velocity is also seen in process reversal. The relative contribution of the two flows
modules on offshore platforms. The duration of this (“expansion” and “contraction”) to the drag force experi-
wind depends on the level of confinement. In general, enced by an object depends, among other things, on its
the higher the confinement, the longer is the duration relative location to the ignition source and inside the gas
and in cases where there is no wall confinement, e.g. on cloud. If an object is traversed by the flame at an early
an FPSO, the duration tends to be short. stage of an explosion, then flow reversal occurs early,
and the main contribution of drag force is from the “con-
2.2. Gas explosion: forces acting on objects in its traction” flow. Conversely, if the object is far away from
path the ignition source, the main contribution of drag force
is from the “expanding” flow. Because the density of gas
An object in the path of the explosion will experience in the expanding flow is much higher than that in the
a load from the pressure wave and the wind. As can be “contraction” flow, it is possible for a piece of equipment
inferred from the discussion above, the nature of this to experience higher drag forces a distance away from
load depends on its size and shape, its location in a pro- the gas explosion than that located within the expanding
cess module, and the arrangement of walls, deck and flame gas explosion. This is particularly true when we
ceiling within the module. consider situations where a flammable gas cloud does
In an unconfined situation the over-pressure wave not cover the entire volume of a process module.
generated will give rise to translational forces on objects
which are bigger or of comparable dimension to the
length scale of the pressure wave. Small objects, such
as small bore pipework only experience drag loading 3. Drag load on pipework
from wind generated by an explosion. This is because
over-pressure equalizes around the object very quickly, Pipe work is often sited around the periphery of a
and the object does not see significant pressure differen- module in areas that may become vent paths for the
tial across it. escaping gases in the event of an explosion. The drag
Grating is a class of objects which does not fit into load on pipe work is dependant on the velocity of the
the categories of small or large objects; its response to enveloping gas and its density. In the course of the event
a gas explosion changes. At low venting speed, the force the temperature of the air will rise from ambient, at say
across the grating is due to drag. As flow velocity 15°C, to 2000°C in less than one-eighth of a second.
becomes sonic, the flow through the grating would be During this time several tonnes of burnt gas will be
“choked” and significant differential pressure will be set vented from the module attaining some extremely high
up across the grating. velocities. The velocity of the expelled gas may
approach 800 m/sec, close to the speed of sound in
2.3. Gas explosion: flow reversal burnt gas.
The load experienced by an object placed in the fluid
There are two main phases to the production of gas flow can be expressed by Morison’s equation.
flow in a confined offshore module—expansion and con-
traction. In the expansion phase, the burning gas expands F = 0.5·CD·gas·V2·D + ·CM·gas·dV/dt·D2/4
and pushes gas outward towards the vent. After the com-
bustion, excess burnt gas is removed and the remaining where D is the diameter of the pipe (m), CD is the coef-
burnt gas cools rapidly and contracts. This creates a ficient of drag, CM is the coefficient of inertia, V is the
negative pressure which drives a flow in the opposite velocity of the gas relative to the component (m/sec),
direction. In theory, this expansion and contraction oscil- dV/dt is the acceleration of the gas relative to the compo-
lation can continue for some time. In practice, this oscil- nent (m/sec2), gas is the density of the gas (kg/m3) and
lation is damped quickly, so only the initial expansion F is the force on the pipe (N/m).
and contraction are significant. For typical pipe sizes and gas velocities the inertia
In an unconfined situation, e.g. FPSO, flow reversal term contributes less than 1% of the force hence the iner-
R.B. Corr, V.H.Y. Tam / Journal of Loss Prevention in the Process Industries 11 (1998) 43–48 45
tia contribution may be neglected reducing the equation ule where the drag load increases to a peak and then
to the normal form of the drag equation: decreases in a manner similar to the rise in over-press-
ure. However, Fig. 2 shows a case where the unburnt
F = 0.5·gas·D·CD·V2. gas attains a significant velocity before it is expelled
ahead of the explosion creating the second peak.
This can be expressed in terms of the gas temperature as
Drag = 0.5·amb·(283/T)·D·CD·V2 3.1. Design of pipe work and supports
where amb is the density of the air at 283 K and T is Pipe work is normally designed for the operational
the temperature of the burnt gas (K). condition and stresses are checked against a pipe stress
Tam & Simmonds (1990) investigated the velocity code. Extreme conditions may be considered such as
profile of gases venting from a module using an early accidental over-pressurization. These codes do not cater
version of the CFD explosion code Flacs. This early for an accidental event such as explosion loading where
work was used as the basis for calculating the kinetic the only requirement is that the integrity of the pipe be
energy that a range of typical loose or dislodged objects maintained post blast.
picked up by the blast wave might attain and these are Designing pipe work for blast loading is not a straight-
given in Steel Construction Institute (1992). forward task. Pipe supports are often engineered to allow
The Tam and Simmond’s model was based on the fol- for thermal movements using a wide variety of support
lowing idealization: arrangements. On an FPSO the flexing of the vessel has
also to be considered. On sliding supports the blast loads
(a) that the gas was burnt, and could cause the pipes to be dislodged unless some form
(b) that the velocity increased linearly to a peak and of movement stop is fitted.
then decreased linearly with equal rise and fall times On some platforms the flow lines have expansion
Plots of the drag magnitude and direction can be eas- loops ensuring that a significant load will be imparted
ily obtained from a CFD explosion model (e.g. Flacs) to the pipe work regardless of the direction of the blast.
enabling design values to be determined. Following a Pipe supports may not be configured to take the suddenly
larger number of simulations on many modules and applied loads from a blast, in particular if the movement
decks it is now apparent that often in the early stages of stops are engaged.
the explosion, where the flammable gas does not cover If one considers a standard pipe run then the simplest
the entire volume of the module concerned, the venting method of analysis will be to apply the peak drag load
gas is at ambient temperature and although the velocity as a static load; a dynamic load factor of 1.6 may be
is considerably lower than at its peak the higher density applied if it is required to maintain the deformations or
gas can create a peak in the drag load. stresses within prescribed elastic limits.
Fig. 1 shows a drag plot for a typical production mod- However, if the increase and decrease in gas velocity
is approximately linear and assuming that the there are be significant even when confinement is minimal, e.g.
no sudden changes in the gas density, then the load time on an FPSO. The total out of balance load on the vessel
history can be approximated to a triangular distribution can be determined by integrating the pressure distri-
and a standard Biggs approach, as described in Steel bution over the surface of the vessel.
Construction Institute (1992), may be used to estimate For a vessel to collapse is extremely unlikely. The
deformations. This approach allows for plastic defor- most likely failure modes will be either toppling, sliding
mation of the pipe work and supports to be taken into or a combination of the two caused by failure of the
account. holding down bolts or failure of the supporting steel
The final most rigorous approach is to undertake a work. Significant movement of the vessel or its supports
full non-linear structural analysis (e.g. using ABAQUS) may lead to failure of the connecting pipe work or dam-
which can account for the non-linear, elasto-plastic age to the nozzles. In either case the result will be a
dynamic effects with the various support conditions release of the vessel inventory.
modelled. It is likely that such an analysis will demon-
strate much higher capacities than either of the simpler
approaches discussed above. 5. Drag load on grating
The grating was mounted inside the test section of the 6. Conclusions
tunnel. A schematic diagram showing the construction
of the grating used on the BP ETAP platform is shown The gases venting from a hydrocarbon explosion may
in Fig. 3. impose significant loads on pipe work and process
High velocity gases were generated in the tunnel and equipment. This may lead to failure of safety critical sys-
passed through the grating. Gas velocities were meas- tems leading to escalation of the event either through
ured as well as load on the grating. Measurements were release of hydrocarbons or loss of control functions.
repeated and found to be repeatable and a calibration The drag loads, in some cases, are not insignificant
check was also carried out using a 5-cm diameter cylin- and require careful design consideration. It is worth not-
der. ing that significant drag loading can result from a gas
The measurement conditions at the grating was up to explosion on an FPSO, and even with a smaller than full
a Mach number of 0.5 and pressure of just over 1 bar. size gas cloud inside a process module.
Results showed that the drag coefficient of the grating It should be ensured that systems have a sufficiently
increased with Mach number. Below 0.3 Mach the effec- high degree of robustness to resist the loads generated
tive drag coefficient was 2. This rose to 3 at 0.5 Mach. by a wide range of possible events and avoid over-
Our analysis indicated that this figure is unlikely to rise reliance on loads obtained from a limited number of
above 4. simulations. It is unlikely the conditions of an actual
The main direction of the drag force was found to be event would be identical to those of a design scenario.
across the grating. The force along the plane of grating Hence, calculated loads are only an indication of the
was small, about 10% of that along the direction of flow. general levels to be anticipated. The distribution and
This result was as expected. direction of the loads for design should be developed by
The measurements showed that the maximum drag the engineer in conjunction with the explosion modeller.
force on the grating used by the BP ETAP Project was There is very little guidance in place to assist the
given by 0.4 V2. It is important to note that this is detailed design of the safety critical systems and design
specifically for the grating used on the BP ETAP plat- of vessels and pipe work for secondary blast effects is
forms, and is not generally applicable. A more general often not considered within the codes.
methodology for calculating drag loading on grating is A methodology has been presented for calculating the
given in Appendix A and Appendix B. effective load on grated floors.
Fig. 3. Dimensions (mm) of the grating tested. Thickness: frame, 4.0 mm; vertical plates, 4.8 mm; horizontal twisted rods, 6.0 mm. Length (in
flow direction), 30 mm.
48 R.B. Corr, V.H.Y. Tam / Journal of Loss Prevention in the Process Industries 11 (1998) 43–48
Assumptions
Draggrating = 0.5CDAV2