Sie sind auf Seite 1von 8

aTe 6477

Design of the Draugen Topsides for the Effects of


Gas Explosions
G. Cockbain and A. Jermstad, Kvaerner Engineering NS, and D. Stenhouse, NS Norske Shell
Copyright 1990, Offshore Technology Conference
This paper was presented at the 22nd Annual OTC in Houston, Texas, May 7-10, 1990.
This paper was selected for by the OTC Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper,
as 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 Conference or its officers. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The
abstract should contain conspicuous acknowledgment of where and by whom the paper is presented.
ABSTRACT
The paper describes the approach taken in the
design of the Shell Draugen Platform integrated
deck topsides to account for the effects of
possible gas explosions. The principal elements
are:
- problem review and definition
- explosion simulation analyses by the FLACS code
- optimization of topside configuration
- derivation of time dependent design loadings
- application of non-linear dynamic structural
analysis.
INTRODUCTION
Until recently, the effects of a partly confined
gas explosion on a platform structure and internals
have generally been assessed in a rather simplified
manner due principally to inadequate methods for
calculating overpressures in large compartments
with complex geometry. The recent availability of
the advanced numerical tool known as FLACS ' (FLame
ACceleration Simulator) has permitted the designers
of the Draugen Platform topsides to follow a more
systematic and realistic approach to optimizing
layout and establishing time dependent design
loadings.
The explosion pressure loadings calculated for many
compartments are high compared to both normal
operational design loads and to explosion design
loads specified for earlier generations of offshore
platform designs.
To avoid undue conservatism and subsequent economic
consequence, whilst complying with the design
acceptance criteria, special design techniques have
been employed.
References and figures at end of paper.
489
Non-linear dynamic structural analysis has been
carried out for representative models of the
structure in the time domain. Stresses, strains
and deformations are calculated for the structural
steel elements to demonstrate compliance with the
acceptance criteria.
The overall sequence of events related to design
for the effects of gas explosion is illustrated in
Figure 1.
THE DRAUGEN TOPSIDES
The Draugen Platform will be installed in summer
1993 in 250 mwater depth on the Haltenbanken,
offshore Mid-Norway. The 22000t operating weight
topsides will be supported upon a concrete
monotower SUbstructure (gravity base). Oil
production capacity is 110,000 BPD. Associated gas
will be re-injected to the reservoir during the
first six years, after which it will be exported by
pipeline. Up to 10 wells can be drilled within the
single water flooded concrete shaft.
The topside which has been designed as a fully
integrated deck, is shown in Figure 2.
Asystem of four main girders runs in the east-west
longitudinal direction and seven girders in the
north-south transverse direction. The deck is
supported by the concrete monotower at the four
intersections of main girders L200 and L300 with
main girders T300 and T500. Plated bulkhead
construction is used for all seven transverse main
girders and for longitudinal main girders in the
non-hazardous quarters and utility areas west of
the central wellbay and drilling facility areas.
In the process areas (Separation and Compression),
located to the east of the platform, the
longitudinal main girders are of open truss
construction.
DESIGN OF THE DRAUGEN TOPSIDES FOR THE EFFECTS OF GAS EXPLOSIONS 2
Area classification boundaries are compatible with
plated bulkhead locations, thus avoiding additional
fire and blast walls.
The individual decks, within the compartments
formed by the intersecting main girders, are
supported by a system of longitudinal deck beams
spanning between the transverse bUlkheads. In the
hazardous classified areas deck plating is replaced
by floor gratings to permit explosion venting down
towards the sea.
The selection of which compartments would be
considered potentially subject to explosion loading
was made on the basis of whether there was a source
of gas within the compartment, regardless of the
overall probability of the occurence or magnitude
of a leak. This was due to difficulties in
predicting probabilities with sufficient accuracy.
The compartments are shown in Figure 3.
A- Process Area 1, T500 -T600 (Separation)
B- Process Area 2, T600- T700
C- Wellbay Area
D- Below Piperack (Compression)
E - Drilling Mud Tank
F - Metering Area
G- Pig Launcher Area
INITIAL ESTIMATES
The development of the topsides structural and
layout configuration was strongly influenced by
initial estimates of possible explosion
overpressure in the various compartments in which
gas might be present.
In particular the extent of open truss structure in
main girders, and the extent of open floor grating
in place of deck plate, were selected to reduce
overpressure as much as possible.
At this stage simplified methods 2 based on
compartment parameters of volume and venting area
were used. Such methods can be expected to
underestimate maximum overpressures since account
is not taken of the effect of internal obstructions
in increasing flame front turbulence and thereby
combustion rate and pressure build-up.
Furthermore, the vent areas assumed should account
for the inevitable reductions in effective area as
topside detail design progresses.
Venting from one compartment to another is not
permitted.
EXPLOSION VENTING AT OUTER HALLS
Provision of maximum possible vent areas as close
as possible to potential ignition sources is
clearly the design objective in order to reduce
explosion pressures. However, working environment
considerations at this northern location, with
their own implications for platform safety, dictate
a need for wind shielding. Where compartments are
to be naturally ventilated, louvre claddings are
required if weather protection is to be maintained.
490
GTC.6477
Recent experimental research 3 has demonstrated the
inefficiency of conventional blowout walls and
louvre claddings in relieving overpressure
build-up. This is due to both loss of effective
vent opening and to the time lapse before which
blow-out walls are fully open.
If a compartment is to be enclosed, the most
effective means to relieve explosion pressure is by
special relief panels with low mass hinged flaps.
The panels for the Draugen platform are specified,
and tested, to open fully within 40 milliseconds of
the internal pressure reaching 0.05 bar.
The explosion relief panels share the same support
system as the louvre panels, but give better
working environment and explosion relief
characteristics and are therefore used to the full
extent compatible with natural ventilation
requirements. Compartments for which weather
protection is not required are of course left
unclad at the outer walls.
ACCEPTANCE CRITERIA FOR STRUCTURAL DESIGN
The design of platform structures located on the
Norwegian continental shelf must comply with
Norwegian Petroleum Directorate Regulations 4. The
Regulations define an explosion as an accidental
event for which the structure shall be checked for
the limit state of progressive collapse (PLS)
condition.
For this condition design safety factors are
reduced and local damage may be tolerated provided
that progressive escalation of the accidental event
can not thereby occur.
The requirements of the statutory Regulations are
complied with, by the specific criteria adopted:
Structural elements essential to overall
topsides integrity shall not generally be
permitted to reach yield stress nor to fail
buckling stability checks.
Other structural elements, including those
contributing to but not essential for the
overall integrity, may reach yield. Plastic
hinge formation is permitted for compact
sections provided that the steel exhibits good
fracture toughness and ductility properties and
that shear forces can, if necessary, be
redistributed. In the event of these elements
failing stability checks or other criteria the
consequences for overall structural integrity
and for platform safety shall be evaluated
specifi ca11y.
Equipment foundations, process piping systems
firewalls, penetration sealings and passive
fire protection shall remain intact.
An investigation of the reserve strength of the
platform was performed to facilitate identification
of the structural elements essential to overall
integrity.
OTC 6477 G. COCKBAIN, A. JERMSTAD AND D.STENHQUSE 3
The study determined the alternative collapse modes
of the integrated deck following hypothetical
failure of successive primary members due to the
actions of rare loading events.
Rare loading events are events (e.g. earthquakes,
fires and explosions) of intensity greater than
those predicted for the 10E-4 annual probability
abnormal or accidental events for which the design
is checked in the PLS condition.
THE FLACS CODE
The FLACS 1 code is the name of the experimentally
verified advanced computer model for simulations of
deflagration type gas explosions in complex
geometries developed by the Chr. Michelsen
Institute (CMI) of Bergen, Norway.
FLACS (FLame ACceleration Simulator) was developed
in the course of a major seven year, oil industry
funded, experimental and theoretical research
programme into the causes and effects of accidental
gas explosions on offshore structures. Research
study 5 was particularly directed at turbulent
flame acceleration and the interaction between the
containing walls, the internal obstacles and the
combusting flow itself.
The FLACS code solves the three dimensional Navier
Stokes dynamic partial differential equations
amplified to include the effects of turbulence and
chemical reactions. The equations are discretized
using a finite volume technique.
Amajor feature of FLACS is the advanced high
resolution computer graphics user interface known
as CASD (Computer Aided Scenario Design), which
facilitates generation of input data and a wide
range of possible results specifications.
FLACS analyses for the Draugen project were
performed by CMI based on scenario specifications
defined by K v ~ r n e r Engineering and Norske Shell.
SCENARIO SPECIFICATION FOR FLACS ANALYSIS
The input data required for specification of an
explosion scenario for FLACS analysis includes:
specification of compartment boundaries
geometry of the structural elements
arrangement of equipment, pipes, ducts etc.
arrangement and specified porosity of gratings
and louvres at vent openings
opening time and pressure of relief panels
gas cloud composition and location
ignition location
output specifications
The output specification possibilities include:
pressure vs. time plots at specified points
pressure vs. time plots, averaged over
specified entire walls and floors.
drag pressure vs. time plots at specified points
impulse curves for local pressure and local
drag pressure (integral of pressure-time)
491
cross-sectional field plots at specific time
steps showing:
- shaded contour overpressure
distribution
- gas flow velocity and direction
vectors
- combustion products distribution,
showing flame development
video presentation of explosion dynamics and
pressure development.
Drag pressure is determined as a function of gas
mixture density and flow velocity for the specific
location and time step.
Figure 4 shows the average pressure-time plot for
wall T500 of process area 1. Figure 5 shows field
plots of overpressure and velocity vectors for a
section through process area 1 at the instant of
maximum overpressure.
GAS QOHPOSITION AND IGNITION
The Draugen design explosion scenarios have been
based upon the assumption of a stoichiometric
gas-air mixture filling the compartment prior to
ignition. The gas compositions for the FLACS
analyses have been specified according to the
primary source within the area. These are given in
Table 1.
The severity of explosion effects was shown to be
highly sensitive to the location of the ignition
point. Ignition locations for design cases were
selected as the "worst reasonable" assumed location
based on equipment or human activity within the
area. Alternative locations were checked in FLACS
analyses to confirm the selection.
OPTIMIZATION OF QOHPARTMENT CONFIGURATIONS
FLACS analysis of the specified base case scenario
was carried out for each compartment. Subsequent
sensitivity analyses were then performed with
revised parameters to develop vent area and
compartment configurations that minimized
overpressures. High priority was given within
the project organization to adapting the topsides
detail design to conform with the recommendations
of the explosion study optimizations.
The timing of these analyses was important since
the input was based upon a high level of
engineering detail and yet the possibility to
change the design was still available.
This process established the arrangement of relief
panels, louvres and unclad openings in walls, and
of floor gratings in decks. Relief panels were
specified along the outside of the main
longitudinal truss girders L100 and L400 and around
three sides of the area under the piperack.
For the drilling mud tank compartment significant
overpressure reductions were achieved through
reduction of compartment volume and complexity by
locally re-arranging positions of gas-tight fire
walls and decks.
pESIGN FOR THE DRAUGEN TOPSIDES FOR THE EFFECTS OF GAS EXPLOSIONS
4
For the central we11bay area. which vents primarily
upwards. additional lower deck vent slots were
introduced to vent out to open air below the
topsides. The permissible footprint area of the
drilling rig above was also limited to improve the
venting upwards.
Explosion overpressures in process area 2
(T600-T700) were further reduced by introducing
explosion relief panels at openings in the T700
east side outer bulkhead.
DERIVATION OF DESIGN LOADINGS
Design loadings in the form of pressure-time
diagrams were developed for the boundary surfaces
of each compartment. The shapes of the diagrams
represent a simplification of the FLACS average
pressure vs. time curves to simple triangular or
composite triangular shapes. The diagrams are
presented in Figure 6 and Table 2.
Peak pressure values for design loading diagrams
were assessed from FLACS pressure contour field
plots as the highest pressure acting upon a
significant part of the surface. The size of this
area was judged according to the structural system
and was typically the square of the smallest of the
compartment length. width or height dimensions.
Design loading diagrams of "drag pressure" vs. time
and vs. position were developed for cross-sections
through compartments. Gas/air mixture flow
velocities. and thus drag loads. are most
significant at the open faces of the compartments.
and decay rapidly towards the interior. This is
illustrated by the drag load diagram for the
mid-height zone in process area 1. shown in
Figure 7. A shape factor or drag coefficient of
0.6 is applied when determining loads on smooth
surface piping.
ANALYSIS OF STRUCTURAL RESPONSE
Non-linear dynamic structural analysis in the time
domain. using the program ABAQUS 6. was performed
for representative structural models of the
compartments. Due to the structural system of the
Draugen deck it was in most cases possible to limit
the explosion analysis model to 2-dimensiona1
frames with multiple beam element members
representing deck plating and beams and the
supporting vertical girders in the main transverse
plated bulkheads. Such frames are representative
of a 3.5 mwide longitUdinal strip of the structure
at locations between the four main longitudinal
girders. In some cases more complex analysis
models were required. Compatibility of boundary
condition stiffnesses and deformations. compared
with the large scale global structural analysis
model. was accounted for. Figure 8 shows a
characteristic section of process area 1 and the
corresponding analysis model.
492
OTC 6477
Astatic and a dynamic analysis was performed in
each case. The dynamic analysis comprised an eigen
frequency analysis followed by a non-linear
analysis in which the appropriate loading diagram
was applied in simulated real time.
The time step length was typically 0.002 sec.
The steel was simulated as an ideal elastic-plastic
material with a horizontal stress-strain diagram
from the yield stress of 420 MPa (61 Ksi).
The beam elements used allowed independent tracking
of stresses and strains at thirteen points in the
element cross-section.
Dynamic amplification factors (OAF) were calculated
for each deck and bulkhead based on comparison of
results from static and dynamic analyses.
OAF values ranged typically between 0.8 and 1.4.
Compliance with the acceptance criteria was checked
either by direct review of the dynamic analysis
results or by application of the appropriate OAF
value in a quasi-static analysis or calculation.
Where necessary. member sizes and connection
details were revised.
For the framework shown in Figure 8 the critical
location was in element 30. corresponding to the
connection of the deck beam lower flange to the
vertical bulkhead. Constructional and welding
details for these locations were therefore
developed accordingly.
DESIGN OF PIPING AND EQUIPMENT FOR EXPLOSION LOADING
Critical hydrocarbon containing piping and
equipment. and their supporting structures. are
designed accounting for the time dependent drag
loading diagrams. Routing of hydrocarbon pipes in
high drag zones is avoided where possible.
Deflections of structural decks due to explosion
pressure can also induce forced deformations of
piping systems, which can be significant
particularly for nozzle connections to pressure
vessels.
INDEPENDENT DESIGN REVIEH
An independent review of the project design
approach was performed by Shell Research Ltd.
Thornton, U.K. assessing overpressure trends and
contributing with advice on specific problem areas.
OTC6477 G COCKBAIN A. JERMSTAD AND D. STENHOUSE 5
CONCLUSION
It is hoped that the design approach presented in
this paper will be of interest and relevance to a
wide audience amongst those responsible for
planning and designing offshore platforms.
The experience of the Draugen project has shown
that is particularly important to give high
priority to explosion design aspects of the
topsides development at an early stage, and that
use of advanced explosion simulation computer
models has become a necessary step in designing for
platform safety.
REFERENCES
1. J.R. Bakke et al.: "Practical Applications
of Advanced Gas Explosion Research. FLACS-
APredictive Tool".
6th Int. Symposium Loss Prevention, Process
Industries, Oslo, June 1989.
2. Det Norske Veritas: Veritas Offshore Standard
D204- "Design against Accidental Loads",1987.
3. Chr. Michelsen Institute: Report No. 25210-1,
"Explosion Testing and Computer Simulations of
Explosion Relief Panels", June 1988.
(Report available through VVS Stord A/S,
Stord, Norway).
4. Norwegian Petroleum Directorate: "Regulation
for Structural Design of Load-bearing
Structures Intended for Exploitation of
Petroleum Resources", 1984.
5. B.M Hjertager, K. Fuhre and M. Bj0rkhaug:
"Gas Explosion Experiment in 1:33 and 1:5 Scale
Offshore Separator and Compressor Modules using
Stoichiometric Homogenous Fuel/Air Clouds".
J. Loss Prevo Process Ind., Vol 1, October 1988.
6. Hibbit, Karlsson & S0rensen, Inc. : "ABAQUS
Ver. 4.6 Theory Manual", 1987.
Table 1. Gas Compositions for FLACS Analyses
COMPARTMENT r... rnmnnoiti on
% Methane %Propane %Butane
A Process Area 1 20 30 50
B Process Area 2 20 30 50
C Well bay Area 50 20 30
D Below Piperack 50 20 30
E Drilling Mud Tank 50 20 30
F Meteri ng Area 20 30 50
G Pig Launcher Area 72 18 10
Table 2. Summary of Design Loadings on Boundary Surfaces
Pressure Peak Load Time (s)
Compartment Peak Value Diagram
(bar) Type tl t2 t3
iA Process Area 1 AP1 0.3 1 0.11 0.05 0.08
c,P2 1.0
B Process Area 2 C,P1 0.5 2 0.18
C We" bay Area C,P1 1.5 2 0.12
D Below Piperack D,P1 0.05
1 0.22 0.04 0.07
D,P2 0.3
E Drilling Mud Tank D,P1 0.5 2 0.21
F Meteri ng Area D,P1 0.8 2 0.22
G Pig Launcher Area D,P1 0.2
3 0.06 0.04
b.P2
0.27
0.05
493
PLATFORM CONCEPT
INTEGRATED DECK
SUPPORTED ON MONOCOLUMN
SUBSTRUCTURE
...
... .. ..
STRUCTURAL DESIGN
FUNCTIONAL REQUIREMENTS SAFETY REQUIREMENTS
REQUIREMENTS
- BASIC STRUCTURAL SYSTEM - EQUIPMENT AND FACILITIES LAYOUT - .AREA CLASSIFICATION/FIRE WALLS
- ACCEPTANCE CRITERIA - WORKING ENVIRONMENT
- NATURAL .. MECHANICAL VENTILATION
- WEIGHT OPTIMIZATION CONSIDERATIONS EXPLOSION RESIST.ANCE " VENTING
+t
I

I
PRELIMIN.ARY
I
INITIAL ESTIMATES OF EXPLOSION
STRUCTURAL CALCULATIONS
OVERPRESSURE BY SIMPLE METHODS

I
+T
I
I
PRELIMINARY TOPSIDES STRUCTURAL AND LAYOUT CONFIGURAnON
T
GLOBAL STRUCTURAL
ESTM3L1SH GAS COMPOSITIONS
ANALYSIS OF TOPSIDES
N40 LIKELY IGNITION LOCATIONS
STRUCTURE FOR
FOR EACH COMPARTMENT WITH
OPERATIONAL LOADINGS
HAZ.ARDOUS CLASSIFICATION
ETC.
GENERATE INPUT
",r
FOR "FLACS" .ANALYSES.
I .-
RESERVE STRENGTH
VARY PARAMETERS
GAS EXPLOSION
ASSESSMENT OF SllAULATION
MAIN STRUCTURE - VENT AAEAS ANALYSES FOR EACH
-
AREAS
- VENT LOCATIONS COMP.ARll.lENT BY
- VENT TYPES "FL.ACS" CODE
:

I
LATED EQU EI.lENTS
.. ,.
FOR ELECTRIC.a.L.
INSTALLATIONS BASED
of
AAEA CLASSIFICATION
I
FROZEN CONFIGURATION OF MAIN STRUCTURE,
I
DERIVE TIME DEPENDENT
EXPLOSION VENTING .ARRANGEMENTS NIlO LAYOUT
DESIGN LOADINGS
.----
OF" MAJOR EQUIPMENT AND SERVICE ROUTINGS
10) WALLS AND FLOORS
t
bl DRAG ON PIPES" EQPT.
t
.,.
r
PERFORM NON-LINEAR DYNPMIC I DESIGN HYDROCARBON
STRUCTURAL ANALYSIS FOR
,+
PIPING SYSTEMS ..
ELEVANT PARTS OF EQUIPMENTIVESSELS
...
IREVISE ELEMENT
DRAG
SIZE IF NECESSAAY
t DEVELOP SAFE
SPECIFY EXPLOSION
CONFIRM ACCEPTABILITY OF
I
't
PERFORM LOCAL DESIGN
OPERATING
RELATED REQUIREMENTS
STRUCTURAl. ELEMENTS IN
OF STRUCTURAl. SUPPORTS
PROCEDURES TO
FOR ACCORDANCE WITH
I
REDUCE RISK OF
- EXPL. RELIEF PANELS ACCEPTANCE CRITERIA
FOR DRAG LOADINGS
IGNITION
- LOUVRE WALLS
- FLOOR GRATINGS
- PASSIVE FIRE PROT.
FIG.1 SEQUENCE OF DESIGN EVENTS
494
FIG.2 THE DRAUGEN TOPSIDES
FIG.3 THE INTEGRATED DECK STRUCTURE
495
41.
LfHGTH.
, ., e 1
1.044-
0.611
0.578
0.34':>
0.111
0.00':>
676.2+'"
5<53.53<5
+50.e2g
332.122
225.415

".
10.
1\
10.
I-'IOIRECTIOH OF DRAG
0,2
-..
--- o. ---
ELEMENT 30 ELEMENT 33
f' -;-:
I o.p
,
i
\
\
\
FIG.5 FIELD PLOTS OF PRESSURE AND FLOW
FlG.7 DRAG LOADING DIAGRAM-PROCESS AREA 1

o
496
P (bara) rnin.--O.005 max.=1.161
146
FIG.8 TYPICAL SECTION TIlROUGH PROCESS AREA 1
Type 3
1ft
.,
1-
+J
E
oD
----.-
Il
'"
E
....
'"
Type2
FIG.6 BOUNDARY SURFACE LQADlRl DIAGRAMS
Type 1
1.00
II.M T500 AVERAGE
'L'
0
II.M
e
n.
Q40

QOO
-Q1O
QOO 0.10 Q40 QIll IIJJ 1.00 1.10 1.l0 1.60
TIme (s)
FIG.4 PRESSURE TIME PWT

Das könnte Ihnen auch gefallen