Professional Documents
Culture Documents
Gladys Nalvarte
DNV GL
E: Gladys.nalvarte@dnvgl.com
T: +47 9060 2309
Confidential
Content
Introduction
Objective
What is done
Influence of design parameters on DAL
– Ventilation
– Dispersion
– Explosion
Influence of process parameters on DAL
– ESD and Shutdown
– Ignition Control
Conclusions
Confidential
1.E-02
Target 1_Small
1.E-03 Process area
Exceedence frequency (1/year)
1.E-04
1.E-05
1.E-06
1.E-07
1.E-08
Confidential
0 0.5 1 1.5 2 2.5 3 3.5 4
Pressure (barg)
3 DNV GL © 2014 16 June 2014
DAL at 1.E-4
Objective
Confidential
DNVGL has run ERA tool that can efficiently be used to perform parameter
studies
A sensitivity analysis was performed to quantify the variation in DAL when
each input to the probabilistic analyses was changed at the time.
Two types of areas were analysed : one large process area and one small area
in a floating unit.
ERA performed 112 times
Confidential
CFD
Ventilation
CFD
Dispersion
Risk analysis:
CFD DNV program EXPRESS
Explosion Response surfaces
JIP Ignition model
Monte Carlo simulations
Improve
design
Explosion DAL
Design effects and
recommendations
Confidential
Max m Max initial leak rate kg/s 50 100 300 900 0.54
No. of ign.sources, 3 increase
Ignition Ns 0.5 1 2 3 0.56
places factor
sources
param Increase of T1
Tsd Time to ign.sd. 0.5 1 2 3 0.57
and T2
Confidential
Max m Max initial leak rate kg/s 182.5 365 730 1500 1.12
No. of ign.sources, 3
Ignition Ns increase factor 0.5 1 2 3 1.06
places
sources
param Increase of T1
Tsd Time to ign.sd. 0.5 1 2 3 0.83
and T2
Confidential
Confidential
Confidential
Example for dispersion: find the largest cloud size within the flammable limits
Q9 – Volume of
Fuel the gas cloud
within the
Air flammable limits
Complexity: Physics & Geometry
Confinement,
Congestion,
Leak (Location, Direction, etc.)
Wind (Direction, Speed, etc.)
Poor local
ventilation Good local
ventilation
Confidential
Confidential
Effect
Influence of of the
the filling C1 parameter
faction steepness in the 1.E-4
coefficient onDAL
the DAL at 1.E-
comparison
4/year for
for the small
small process
process area area
Poor local
ventilation
Good local
ventilation
Total Variation
0.7 barg
Confidential
1.E-02
C1=3
C1=10
1.E-03 Good local C1=30
Exceedence frequency (1/year)
ventilation C1=90
Poor local
1.E-04 ventilation
1.E-05
1.E-06
1.E-07
0 0.5 1 1.5 2 2.5 3 3.5 4
Overpressure (barg)
Confidential
Jet zone
Passive zone
Confidential
E
Vfmax/V
=
0.13 Vfmax/V=
0.06
Vfmax/V=
NW 0.15
Vfmax/V=
W 0.16
Confidential
Confidential
More global effect that will produce the biggest gas cloud
1.6
Poor
1.4
ventilation
Good
1.2
ventilation
1
DAL (barg)
0.8
0.6 High
variation ~ 1
0.4
barg
0.2
0
0.1 0.2 0.3 0.5
Average A+B
Confidential
Good Poor
ventilation ventilation
Confidential
Confidential
0.5
Confinement – Amount of walls
0
and decks 0 200 400 600 800 1000
Cloud size, Vf (m3)
– Av/A, confinement parameter,
Open area per total surface 3
Plot 1, Av/A is parameter
area 2.5
Av/A = 0.10
Av/A = 0.12
2
Pressure (barg)
Av/A = 0.20
Distance - cloud size 1.5
1
Gas type and concentration
0.5
0
Confidential 0 200 400 600 800 1000
Cloud size, Vf (m3)
DNV GL © 2014 16 June 2014
22
Confidential
Confidential
11
Response surface, f1
10
7
P (barg)
3
6 E p1
5
f1 e
4
3
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Confidential
Ec/Emax Filling
Pressure linear
increase
Kp
Pressure
exponential
increase
P1
Confidential
Lower
congestion
and
Higher
confinement
congestion
and
confinement
Confidential
Higher
Low congestion/ congestion/
confinement confinement
Confidential
Confidential
Leak profiles
Confidential
Confidential
Confidential
Confidential
Conclusion (continuation)
Confidential
Gladys Nalvarte
Gladys.nalvarte@dnvgl.com
+47 9060 2309
www.dnvgl.com
Confidential
Q: You seem to have characterised explosions via their peak pressures. Did you also
consider their durations in your analysis?
A: No. We only considered the overpressures on the target area in the presented study.
However, it is possible to establish a correlation between overpressure and duration
in the analysis.
Q: I understand that the presented sensitivity study was carried out by changing one
parameter at the time whilst keeping the others unchanged. Is it correct to assume
that the parameters described in your presentation are independent from each other
and that changing one does not affect others?
A: We do know that this is not entirely the case and that some parameters are not
independent. The type of analysis described in my presentation is however very
efficient at the preliminary design stage for identifying the parameters which influence
the risk the most.
A: The leak durations to be considered will depend on type of installation. The values
shown in my presentation corresponded to the small and large process areas
investigated in this case.
Q: How do you account for the non-uniform mixing of the gas cloud with air in your
explosion analysis?
Q: You showed average values in your presentation. How was this “averaging”
carried out?