You are on page 1of 6

OTC 14134

Investigation of Gas Dispersion and Explosions in Offshore Modules


D.M. Johnson, Advantica Technology Inc., R.P. Cleaver, Advantica Technology Inc., J.S. Puttock, Shell Global Solutions,
C.J.M. Van Wingerden, GexCon.
Copyright 2002, Offshore Technology Conference
This paper was prepared for presentation at the 2002 Offshore Technology Conference held in
Houston, Texas U.S.A., 69 May 2002.
This paper was selected for presentation by the OTC Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Offshore Technology Conference and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Offshore Technology Conference or its officers. Electronic reproduction,
distribution, or storage of any part of this paper for commercial purposes without the written
consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print
is restricted to an abstract of not more than 300 words; illustrations may not be copied. The
abstract must contain conspicuous acknowledgment of where and by whom the paper was
presented.

Abstract
There has been considerable progress over the last decade in
validating models to predict the loading resulting from an
explosion in an offshore module. However, much of the
experimental work that provided the validation of the models
has involved using idealized gas clouds, generally natural gas
or methane at a single concentration throughout the whole
volume. This paper describes an experimental project and
modeling work undertaken to provide a more realistic
representation of the risks posed by gas explosions in offshore
platforms. The project involved laboratory, medium and large
scale experiments, examining the effects of realistic fuel
mixtures, realistic gas release conditions and nonhomogeneous gas clouds filling only a portion of the module.
Comparison with gas dispersion and explosion models was
also carried out.
Background
The Blast & Fire Engineering Project for Topside Structures
[1] was initiated in May 1990 to study the special explosion
and fire hazards facing offshore operators. The objective of
Phase 1 of this joint industry project (JIP) was to provide
interim guidance to designers and operators to reduce the
potential for a major disaster offshore as happened on the
Piper alpha platform in 1988. In addition to providing this
interim guidance, this project recognised that there was a need
to obtain data on these hazards from full-scale experiments.
Therefore, as part of the follow-on Phase 2 project, a series of
experiments was undertaken to obtain quantitative data on the
explosion hazard in full-scale geometries truly representative
of the offshore environment [2].
The results of this project indicated that high explosion
overpressures could be generated and that water sprays

activated prior to ignition could significantly reduce these


overpressures. As a consequence of these findings, the UK
Health & Safety Executive (HSE) commissioned a further
project, known as Phase 3A [3], in which experiments were
undertaken to identify methods of reducing the severity
of explosions.
The Phase 2 and Phase 3A projects showed that it is
generally not practical to design an offshore platform against
the 'worst case' explosion. Therefore, a risk based approach
for design against explosion hazards is needed in which the
risks are reduced to as low as reasonably practicable. In
taking a risk-based approach, it is important that the full
spectrum of possibilities is adequately understood. This
includes scenarios involving ignition of a non-homogeneous
gas cloud occupying only part of the module. The Phase 2 and
3A projects had, however, been conducted with homogeneous
gas clouds throughout the full module, and in most cases the
mixture had been at a stoichiometric concentration. This was
appropriate at the time, as it was necessary to understand of
the scope of the explosion hazard and to provide well defined
experiments against which explosion models could be
validated. However, there was a clear need to investigate
more realistic conditions.
As part of this investigation, a JIP was carried out to study
the dispersion of high pressure gas releases in offshore
modules [4]. This revealed that the releases produce a variety
of different types of dispersion or gas accumulation behaviour,
depending on particular combinations of perimeter
confinement, congestion and release direction and ambient
weather. However, by no means all of the releases resulted in
a uniform concentration distribution of the type studied
previously in the explosion projects.
As a result, a project, known as Phase 3B, has been
undertaken in order to gain information on the explosion load
generated in more realistic conditions. The project has been
sponsored by BG Group, BP, Enterprise, ExxonMobil, HSE,
Institute of Petroleum, Marathon, Norsk Hydro, Shell, Statoil,
Talisman, TotalFinaElf.
The project was managed by
Advantica Technology and conducted by Advantica
Technology, Shell Global Solutions and GexCon. This paper
provides a summary of the work carried out within the project.

D.M. JOHNSON, R.P. CLEAVER, J.S. PUTTOCK AND C.J.M. VAN WINGERDEN

Project Scope
The project comprised experimental and modelling studies,
with the experimental work being carried out at laboratory,
medium and large scale. The main elements of the work are
detailed below.
Laboratory Scale Experiments. Most of the experimental
work carried out prior to the Phase 3B project had used natural
gas and methane fuel-air mixes. However, there are many
different gas compositions within the process inventories on
offshore platforms. These compositions may have significant
higher hydrocarbon and/or inert components. Thus, as part of
providing data that would broaden the understanding of more
realistic scenarios, laboratory scale tests were carried out in a
Duct Rig at the Shell Global Solutions Thornton site. The
experiments had the objective of estimating the fuel-specific
turbulent combustion characteristics of a range of hydrocarbon
mixtures, some of which were diluted with nitrogen or carbon
dioxide. This information provides support for explosion
modelling and can also be used to determine in general when a
composition variation might have a significant influence on
the resultant explosion load in realistic conditions. The
mixtures were selected to represent the type of natural gas
mixtures likely to be encountered in practice.
Medium and Large Scale Experiments. The primary
objective of the project was to conduct experiments to assess
the explosion load produced by realistic gas clouds. As the
effects produced in an explosion are scale dependant [5], the
best form of validation is against data from experiments
at a scale comparable to an offshore module. However, the
cost of conducting such experiments prohibits extensive
parameter variation.
It was therefore decided that the project should include a
mix of large scale and medium scale experiments, with the
medium scale tests providing a much wider parameter
variation than the large scale tests.
The medium-scale explosion experiments were conducted
at the GexCon Sotra test site in a test rig measuring 8m long

OTC 14134

and 2.5m square, shown in Figure 1.


The large scale explosion experiments were conducted at
the Advantica Spadeadam test site in a test rig measuring 28m
long, 12m wide and 8m high, shown in Figure 2. Both
experimental programmes involved three main types of
experiment, as described below.

Figure 2 Large Scale Test Rig

In conducting the experiments, the following


measurements were made:
Overpressures inside and outside the test rig
Flame arrival times
Gas concentrations
Weather conditions
High pressure gas release rate (realistic release
experiments only)
Base Case Experiments. These tests involved
homogeneous, flammable clouds that filled the entire rig.
They were conducted with the same boundary confinement,
equipment and ignition locations used in subsequent
experiments in more realistic conditions, enabling a
comparison to be made between the explosion load generated
in the more idealised conditions used in previous experimental
studies and that produced in the more realistic conditions.
Partial Fill Experiments. These experiments involved
homogeneous, flammable clouds occupying only a part of the
rig. Such experiments offered an intermediary step between
the fully realistic conditions, where the gas cloud is produced
by a high pressure gas release, and the full rig fill experiments
in Phase 2 and 3A. In these partial fill experiments the
experimental conditions, such as the concentration and
location of the gas cloud, are well defined. This enables
comparison with model predictions in an incremental step
towards fully realistic conditions.

Figure 1 Medium Scale Test Rig

Realistic Release Experiments. These experiments involved


ignition of clouds generated by high pressure gas releases
within the test rig. Gas build-up prior to ignition and the
explosion load produced following ignition of the gas cloud
were both measured in the experiments. The majority of the

OTC 14134

INVESTIGATION OF GAS DISPERSION AND EXPLOSIONS IN OFFSHORE MODULES

experiments were conducted with natural ventilation through


the perimeter of the test rig. A small number of medium scale
experiments were also conducted with forced ventilation
provided through a polythene chute attached to one end of the
test rig. These offer an intermediary step towards fully
realistic conditions, where the ventilation is more variable.
Modelling Studies. During the experimental programme,
GexCon used the FLACS explosion model to predict the
results of the explosion experiments. These predictions were
used to evaluate the ability of CFD models to predict partial
fill and realistic explosion scenarios.
Although the primary aim of the project was to investigate
the explosion load, the realistic experiments provided a wealth
of gas build-up data. This data has been compared to model
predictions by:
GexCon using the FLACS code.
Shell Global Solutions using the random walk DICE
model.
Advantica, providing a comparison with zonal and
free jet model predictions.
Structural Response. The Phase 3B project also included a
programme of work to examine structural response. The
structural response test programme and the evaluation of the
results have been reported separately by Shell Global
Solutions [6].
The experimental programmes are described below, including
details of the experimental arrangements, the parameter
variations and the results obtained.
Laboratory Scale Experiments
The flame acceleration duct rig at Thornton has been
developed as an experimental facility to investigate the
turbulent flame acceleration through obstacle fields [7, 8] in
partially confined explosions. Previous studies [9, 10] have
investigated turbulent burning characteristics of methane,
propane and ethylene when premixed with air at near
stoichiometric concentrations. Steady turbulent flame
propagation with constant flame velocity over a considerable
length (> 50%) of the duct was observed when the roof of the
duct was opened to allow venting of the combustion products.
As discussed by Taylor [11] the minimum amount of roof
venting required for steady turbulent flame propagation is
dependent on the mixture reactivity. Previous work carried out
at Thornton had shown that steady flame speeds were
achieved for methane, propane, and butane mixtures for a
wide range of roof openings (10%, 30%, 70%, and 100%), and
two grid spacings of 0.1 m, and 0.2 m.
The measured steady state flame speed characterises the
turbulent combustion of the gas mixture for various levels of
turbulence intensity. Quantification of turbulent burning
velocity is critical in the prediction of gas explosions as, in
simple terms, overpressure tends to be proportional to burning

velocity squared. The principal parameters which determine


turbulent burning velocity for a fuel at a given turbulence level
are laminar burning velocity, and sensitivity of the flame to
turbulent stretching. The latter is characterised by a Markstein
number. By comparison of the results for a tested fuel mixture
with results for pure methane, and propane, Shell Global
Solutions has developed a technique for determining these
parameters from the rig data. The method provides an insight
into the fundamental processes that underlie gas explosions
and enables a diverse range of mixtures to be studied at
relatively low cost.
Experimental Details. The technique of operation of the
duct rig has been described in detail in the reports cited in the
introduction and a relatively brief description is given here.
The flame acceleration apparatus at Thornton consists of a
duct of dimensions 2000mm x 200mm x 100mm which can be
filled with a flammable gas mixture. Ignition occurs at one end
of the duct and the flame progresses toward the vent at the
opposite end. The duct can contain up to 19 rows of obstacles
of varying drag and blockages that generate turbulence as the
flame passes them. The roof of the duct can vary between
being 90% enclosed to being completely open. The flame
speed in the duct reaches a steady state as the tendency for the
flame to accelerate as it passes the obstacles is offset against
the flow of volume out of the roof.
In the experiments, ion probes connected to a microlink
high speed data capture device measured the time of arrival of
the flame. The external pressure at the exit plane of the duct
rig was measured using a hydrophone.
The experimental programme consisted of 180
experiments comprising different sets of conditions with each
test repeated. Table 1 summarises the gas mixtures
investigated. Tests with these 16 different gas mixtures were
conducted with grid spacings of 0.1 m, and 0.2 m under roof
openings of 0.1, 0.3, 0.7, and 1.0. The tests with roof openings
of 0.7, and 1.0 were carried out in addition to the originally
planned experimental program in order to predict the laminar
burning velocities more accurately.
Table 1 Fuel Compositions and Stoichiometries
Investigated in Duct Experiments
Gas Composition (volume %)
Methane 50%, Ethane 50%
Methane 75%, Ethane 25%
Methane 50%, Propane 50%
Methane 75%, Propane 25%
Methane 45%, Ethane 15%, Nitrogen 40%
Methane 45%, Ethane 15%, Carbon Dioxide
40%
Methane 60%, Ethane 20%, Carbon Dioxide
20%
Methane 50%, Ethane 25%, Propane 25%

Stoichiometry
1.0
0.8, 1.0, 1.3
1.0
0.8, 1.0, 1.3
0.8, 1.0, 1.3
1.0
1.0
0.8, 1.0, 1.3

Though the results derived in the laboratory experiments


are principally for use in explosion models, the effect of the
various changes have been demonstrated.

D.M. JOHNSON, R.P. CLEAVER, J.S. PUTTOCK AND C.J.M. VAN WINGERDEN

Medium Scale Experiments


Two perimeter confinements were used in the project. Most of
the experiments were performed with two long walls and a
roof providing vent openings only at the ends, total vent area
12.5m2. The second geometry was more open. The two
central wall elements (of four wall elements) on one of the
long walls were removed increasing the vent area by another
10m2.
In the base case experiments and in the partial fill
experiments (where required) the vent openings on the test rig
were covered by thin polyethylene sheet to enable gas filling
of the rig. The plastic was fixed to the rig using a remotely
operated plastic release mechanism. This enabled the plastic to
be released typically one second before ignition. In the
realistic release experiments, the vent openings were not
covered by plastic sheet. However, in some experiments a
plastic sheet was used to establish a funnel between a
ventilation fan and the module to generate a well-defined wind
field inside the module. This plastic funnel was not removed
prior to ignition.
In order to be able to retain flammable clouds partially
filling the facility two roller blinds were installed. The roller
blinds consisted of a blind running from the bottom deck to
the roof of the module. As the roller was mounted on top of
the module a narrow slit was introduced in the roof for the
blind to be wound around the roller. The roller was operated
by an electric motor in order to remove the blinds just prior
to ignition.
The gas mixtures were ignited by means of an electric
spark. Six different locations were used for the ignition source
during the test programme.
Overall, 25 base case, 25 partial fills and 72 realistic
release tests were carried out.
Base Case Experiments. Fifteen tests with stoichiometric
mixtures and ten tests with non-stoichiometric mixtures were
performed. Methane was used as the fuel gas in all tests.
For the non-stoichiometric mixtures, three gas
concentrations were used (nominally 7.75, 12 and 12.9%).
Partial Fill Experiments. The positions of the roller blinds
were approximately 2.25m and 3.75m from one end of the
module. The roller blinds allowed cloud sizes of 18%, 28%
and 46% of the total volume of the module to be studied.
Realistic Release Experiments.
The majority of the
experiments (58) were performed in the more confined
configuration. The remaining experiments were conducted
using the more open confinement configuration.
The 58 experiments in the more confined module included
8 release positions/orientations, 3 release rates (0.05, 0.1 and
0.2 kgs-1), and 2 ventilation conditions (weak wind (<1 m/s)
and stronger wind (>1 m/s)). In 11 of the 58 experiments
controlled forced ventilation was used instead of naturally
generated ventilation. This forced ventilation was achieved
using the polythene funnel and fan described earlier.
The 14 experiments in the more open module include 5

OTC 14134

release positions/orientations and the same release rates and


natural ventilation conditions as above.
Large Scale Experiments
The test rig used for the large scale experiments was
constructed in 1994 for the explosion experiments conducted
as part of the Blast and Fire Engineering Project for Topside
Structures, Phase 2. It is an all steel construction extending up
to 28m long, 12m wide and 8m high, as shown in Figure 2.
Within the rig, large and small steel obstacles are positioned to
simulate typical process plant and pipework that would be
found on an offshore installation. The test rig structure is
based on a strong bolted framework, with beams on 4m
centres, onto which panels can be fitted to provide the
perimeter confinement conditions required in the test
programme. A mezzanine deck, consisting of a steel support
frame covered with serrated open bar grating, is located at
mid-height throughout all of the test rig and was designed to
be similar to those found in typical offshore modules.
Two confinement configurations were used in the
experimental programme, denoted C1 and C2. For all of the
experiments, the roof and one 28m long face were completely
confined. For confinement configuration C1, the other 28m
long face was about 3/5th confined, with the open areas in the
central region of the wall. The two 12m long faces were open.
For confinement configuration C2, one 28m long face was
completely open and the two 12m long faces were
fully confined.
Two ignition positions were used in the C1 confinement
configuration and one ignition position was used in the
C2 configuration.
For the base case and the partial fill experiments, the
flammable mixture was retained within the test rig by
polythene sheet. In the partial fill experiments, in order to
prevent the polythene sheet from influencing the experiment,
it was cut free using low energy detonating chord immediately
prior to ignition.
Base Case Experiments. Four base case experiments were
conducted, three in the C1 configuration and one in the C2
configuration. Three base case experiments were required for
confinement configuration C1 as changes were made to the
equipment within the test rig between the partial fill and
realistic release experiments. All of the base case experiments
were conducted with a nominally stoichiometric natural gasair mixture.
Partial Fill Experiments. The partial fill experiments were
all conducted using the C1 confinement configuration
and a single ignition location. All of the partial fill
experiments were conducted with nominally stoichiometric
natural gas-air mixtures.
Five partial fill experiments were conducted at large scale.
The cloud sizes used were 10%, 19% and 43% of the full test
rig volume. For the 19% and 43% clouds, experiments were
conducted with the cloud in a two different locations for each
cloud size.

OTC 14134

INVESTIGATION OF GAS DISPERSION AND EXPLOSIONS IN OFFSHORE MODULES

Realistic Release Experiments. For experiments involving a


high pressure gas release, the gas flow rate was controlled to
be nominally constant for each experiment. The release took
place through an orifice of diameter 32.5mm or 43mm,
pointing in the direction of one of the three coordinate axes of
the test rig. Release rates of between 2.1 and 11.7 kgs-1 were
used in the experiments. In all twenty experiments were
carried out with a high pressure gas release.
Summary of Conclusions
The work carried out provided a substantial amount of
information with detailed conclusions that cannot be included
within the limits of the space available to this paper. As a
consequence, only the primary conclusions of the study are
detailed in the sub-sections below.
Laboratory Experiments. In addition to providing detailed
information suitable for explosion model development, the
experiments also indicate that:
Carbon Dioxide had a greater effect in reducing the
overpressures than Nitrogen.
The presence of inert components such as Nitrogen
and Carbon Dioxide in natural gas are only likely to have a
significant effect in reducing explosion overpressures if their
concentration in the natural gas is in excess of 20%.
The presence of higher hydrocarbon components,
such as ethane and propane will have the effect of increasing
the explosion overpressures. The influence of propane can be
expected to be greater in this respect than ethane. In both
cases, increases in overpressures were observed when the
concentration of ethane or propane in the fuel was 25%.
It has been demonstrated that the addition of a small
amount of propane to methane does not make the mixture as
reactive as propane. The overpressures produced by adding,
for example, 25% propane to methane is well below that for
pure propane.
Medium and Large Scale Experiments.
The main
conclusions from the medium and large scale partial fill and
realistic release experiments are detailed below.
Partial Fill Experiments
The partial fill experiments showed that, for the fill
ratios studied, overpressures reduced as the size of the gas
cloud was reduced.
In the medium scale partial fill experiments where
non-stoichiometric fuel concentrations were used, the
overpressures were significantly lower than for equivalent
tests with stoichiometric mixtures. In addition, in tests with
rich mixtures, higher pressures were produced than in
equivalent experiments with lean mixtures. However, in the
tests where the cloud was filling the entire module the lean
mixtures gave rise to higher pressures than the rich mixtures.
Examination of the test configuration shows that in the partial
fill experiments, the unburnt gas was able to mix with air in
the test rig, causing rich mixtures to become more reactive and

lean mixtures to become less reactive. The effect this would


have on overpressure generation is consistent with the
observations. Only stoichiometric gas clouds were used in the
large scale partial fill tests.
Realistic Release Experiments
Depending on release and experimental configuration
and the weather conditions, realistic releases give gas clouds
that can be fairly uniform over parts of the rig or can have
large concentration gradients within them.
Though considerable variation in flammable cloud
size and concentration was observed in the experiments, under
some release and weather conditions, significant portions of
the module were filled with near stoichiometric natural gas-air
mixture. These conditions were generally produced by
impacting releases where the confinement, release rate and
weather conditions resulted in recirculation and gradual
accumulation of the natural gas-air mixture within the rig.
The overpressures generated in the realistic release
experiments were in general lower than those produced in
equivalent base case experiments. However, in experiments
where there were significant proportions of the test rig was
filled with near stoichiometric mixture, the overpressures
produced were comparable to or exceeded those generated in
the base case experiments.
Regions containing gas that is rich or above the upper
flammable limit tend to be trapped near walls or the roof
where the degree of ventilation is low. For the ignition
positions studied; the rich mixture was effectively trappedby
the solid boundaries as the flame front approached from all
sides. As a consequence, these rich regions were not diluted
as the explosion progressed and therefore do not seem to
contribute to the generation of overpressure. This differs from
the findings from the partial fill experiments, where mixing of
air with the rich mixture was possible.
The ignitability of the mixture depended on the
nature of the gas cloud. Where gas clouds contained a
significant region of approximately uniform mixture, ignition
generally occurred at the first attempt. For experiments where
there was a significant variation in the gas concentration
throughout the test rig, for example in regions of high shear in
the jet-like flows generated by the release, ignition often
occurred only after several attempts.
The jet dispersion, concentration field, flame
development and overpressure data were all self-consistent in
each experiment. In addition, the jet dispersion and early
stages of overpressure generation (for example, before
interaction with the perimeter walls) could be explained by
'conventional' understanding developed from previous work.
It is possible to use a normalised fill and normalised
overpressure curve to give an upper bound to the likely level
of overpressure. The scatter in the figures presenting the
normalised data appears to be related to the initial turbulence
in the realistic release experiments tending to enhance
overpressures and the lower pressures generated by some
experiments in which rich mixtures or mixture above the

D.M. JOHNSON, R.P. CLEAVER, J.S. PUTTOCK AND C.J.M. VAN WINGERDEN

upper flammable limit was present.


Modelling Studies
Explosion Load
Comparison of CFD predictions with the results
obtained from the large scale partial fill experiments showed
that there was a need to ensure that the smaller cloud was
sufficiently resolved. However, when this was done, the
agreement between experiments and predictions was good.
The CFD explosion predictions carried out of the
realistic release experiments comprised linked simulations of
the gas build-up and the explosion process. As a consequence,
the variation in the explosion predictions was influenced by
differences between the predicted and actual gas build-up.
However, in general the comparison between predictions and
measurements was good.
Gas Build-Up
Good correlation between experiments and CFD
simulations was seen for many of the large scale release
scenarios.
At medium-scale many of the experiments are
considered to have a limited value for detailed comparison
with predictive models as the variations in wind condition
disturbed the mixture accumulation.
Comparison of predictions from a Random Walk
dispersion model with the results from the large scale
experiments showed that the general trend of the results is
well captured and the accuracy is close to that of CFD.
A Zonal model correctly predicted the time-scale and
amount of gas accumulation in the experiments where
recirculation took place. A free jet model correctly predicted
the near-field behaviour for releases directed towards an open
face of the rig.
Implications
The experimental programme described in this report provides
a definitive study of the explosion hazards presented by high
pressure gas releases in offshore modules. The results
obtained from the study have a number of clear implications
for explosion hazard management.
Firstly, it is apparent that the overpressures predicted by
considering modules filled with stoichiometric gas clouds
considerably over-estimates the loading produced in many
realistic scenarios. However, it has also been shown that these
'worst-case' conditions can arise, albeit with a low probability.
It is also apparent that these 'worst case' conditions are more
likely to occur when recirculation and accumulation of the
gas-air mixture occurs.
Primary explosion hazard management will probably be
provided by good design and operational practice, such as
improving natural ventilation, providing reliable gas detection
and shutdown systems and limiting the potential for accidental
releases. However, there often remains a need to quantify the
residual explosion risk in order to identify if other
improvement measures are justified or beneficial. This can be

OTC 14134

particularly relevant to new designs, where risk reduction can


more readily be achieved.
This project has shown that the evaluation of explosion
hazards should consider the full spectrum of explosion
scenarios to avoid being overly conservative. It is recognised
that the combination of the gas build-up and explosion
processes is highly complex. However, it has also been shown
that 'detailed' modelling techniques, such as CFD and others,
can be used to provide reasonable estimates of the gas
dispersion and explosion behaviour in realistic scenarios.
Work carried out within the project has shown that simpler
calculation methods, based on correlations, can be proposed,
provided their limitations are recognized.. It has not been
possible to fully explore or validate these techniques within
the project; however, these techniques offer the potential for
simpler explosion hazard management techniques to
be developed.
References
1
2
3
4

7
8
9
10

STEEL CONSTRUCTION INSTITUTE, "Interim Guidance


Notes for the Design and Protection of Topside Structures
Against Explosions and Fire", Report No. SCI-P-112, 1992.
THE STEEL CONSTRUCTION INSTITUTE, "Blast and Fire
Engineering for Topside Structures, Phase 2, Final Summary
Report." SCI Publication Number 253, 1998.
T.Al-Hassan and D.M. Johnson, "Gas Explosions in Large Scale
Offshore Module Geometries: Overpressures, Mitigation and
Repeatability", OMAE 98, Lisbon., May 1998.
R.P. Cleaver, S. Burgess, G.Y. Buss, C. Savvides, V. Tam, S.
Connolly, and R.E. Britter, "Analysis of Gas Build-up from
High Pressure Natural Gas Releases in Naturally Ventilated
Offshore Modules", 8th Annual Conference on Offshore
Installations: Fire and Explosion Engineering, Lords
Conference and Banqueting Centre, London, 30 November,
1999.
W.P.M. Mercx, D.M. Johnson and J. Puttock, "Validation of
Scaling Techniques for Experimental Vapour Cloud Explosion
Investigations", Process Safety Progress, 14, No. 2, 120-130,
April 1995.
P. Bailey, S. Chynoweth, J. Friis, L.A. Louca, T.C. Rhodes,
L.C. Shirvill, "Blast Loading from Gas Explosions onto Nearby
Structures", Offshore Technology Conference 2002, 14129
Houston.
P.H. Taylor, "Fast flames in a vented duct." Intl. Comb. Symp.
21, (1986).
P.H. Taylor and S.J. Bimson, "Flame propagation along a
vented duct containing grids." Intl. Comb. Symp. 22, (1988).
A.T. Cates and S.J. Bimson, "Fuel and obstacle dependence in
pre-mixed transient deflagrations." 13th. ICoDERS, Nagoya,
Japan (1991).
J.S. Puttock, M.R. Yardley and P. Snowdon, "Final Report on
Shell Research contribution to project EMERGE." OT. 96.
40058 (1996).

You might also like