Documente Academic
Documente Profesional
Documente Cultură
Dipartimento di Ingegneria Civile e Industriale Universit di Pisa, largo Lucio Lazzarino, 56126 Pisa, Italy
Dipartimento di Ingegneria Civile, Chimica, Ambientale e dei Materiali, Alma Mater Studiorum Universit di Bologna,
via Terracini n.28, 40131 Bologna, Italy
ar t ic l e i nf o
a b s t r a c t
Article history:
Received 1 August 2013
Received in revised form
14 June 2014
Accepted 17 July 2014
Available online 1 August 2014
Severe accidents may be triggered by the impact of oods on process and storage equipment containing
hazardous substances. The present study analyses the possible damage of horizontal cylindrical
equipment, either operating at atmospheric or at higher pressures. A mechanical damage model was
developed and validated by available literature data on past accidents. Simplied correlations were then
obtained to calculate the critical ooding conditions leading to vessel failure. A fragility model was
proposed for the straightforward assessment of equipment damage probability in the framework of the
quantitative risk assessment of NaTech scenarios triggered by oods. A case-study was discussed to test
the potentialities of the method.
& 2014 Elsevier Ltd. All rights reserved.
Keywords:
NaTech
Flood
Major accident hazard
Hazardous materials release
Storage tanks
Pressure vessels
1. Introduction
Global climate changes may signicantly inuence temperatures and precipitation patterns, increasing the frequency of severe
meteorological events which may lead to ooding [15]. Flooding
may dramatically impact on population among urban areas but it
can also affect industrial facilities. In the latter case, damages to
equipment and structures may lead to the release of hazardous
materials with potential escalation leading to domino effects
[612]. This type of accident scenario is indicated as a NaturalTechnological (NaTech) event. Specic studies [1319] and recent
publications concerning the analysis of industrial accidents
reported in available databases [11,20] highlighted that NaTech
accidents often generate severe consequences. Besides, the severe
events that affected several industrial sites and the Fukushima
power station after the Thoku earthquake and consequent
tsunami in Japan (April 2011) [13,21] showed the potential dramatic consequences of NaTech events.
Therefore, NaTech events may signicantly impact on the risk
prole of an industrial facility. However, the implementation of
NaTech scenarios in conventional Quantitative Risk Assessment
(QRA) studies is a critical task [16,2224]. In particular, the
http://dx.doi.org/10.1016/j.ress.2014.07.016
0951-8320/& 2014 Elsevier Ltd. All rights reserved.
126
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
horizontal cylindrical vessels are mostly damaged due to displacement caused by water drag and/or to oating [7,28].
The present study was devoted to the development of a model
for the vulnerability assessment of horizontal cylindrical process
and storage vessels involved in ood events. In order to evaluate
the resistance of the equipment items considered, a mechanical
model was developed. The model, validated with respect to the
available data obtained from past accident records, was applied to
derive simplied vulnerability functions to calculate vessel failure
probability in ood events. In order to explore the model features
and potentialities some case studies based on actual industrial
lay-outs were analyzed.
Representation of vessel
geometry
Characterization of flood
impact vector
Model validation
Development of
simplified correlations
Evaluation
of vessels vulnerability
Fig. 1. Schematization of the methodological approach adopted for the development of a vulnerability model for horizontal vessels involved in ood events.
In the present study, storage tanks and process vessels consisting of a horizontal cylindrical body with spherical edges were
considered. The vessels, operating at atmospheric pressure or
higher, were assumed as disposed on saddle-type supports, xed
to the ground. The references for the design and features of the
tanks considered in the present study are the API Standard 620
[32] and the ASME Pressure Vessel Code (Sec. VIII of the ASME
Boiler and Pressure Vessel Code [31]).
The schematic representation of these vessels is reported in
Fig. 2a, while the relevant mechanical features considered are
summarized in Table 1. As shown in Fig. 2a, one of the vessel
saddles is assumed to be xed to the ground with a bolt connection, while the other assumed to be only laid on the ground. This
conguration is frequently adopted in the process industry in
order to limit the stress due to steelwork thermal expansion [33].
Table 1
Main features of the vessels included in the database considered in the present study. For the denition of geometrical parameters (dimensions) see Fig. 2. Pd design pressure; ATM operating at atmospheric pressure.
ID
Capacity
3
Dimensions (mm)
(m )
l1
l2
l3
S1
S2
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
ATM
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
2
2
2
2
2
5
5
5
10
10
15
20
20
20
25
25
25
30
30
50
50
50
50
100
100
100
115
150
250
5
5
5
10
10
15
20
20
20
25
25
25
30
30
50
50
50
50
100
100
100
115
150
250
5
5
5
10
10
1600
1300
1000
1600
1200
1800
1500
2000
1900
1700
2200
2300
2400
1900
2100
2800
2700
2500
3200
3200
2800
2750
3200
3800
1600
1300
1000
1600
1200
1800
1500
2000
1900
1700
2200
2300
2400
1900
2100
2800
2700
2500
3200
3200
2800
2750
3200
3800
1600
1300
1000
1600
1200
3000
3500
6100
4500
7700
5500
9700
6000
7200
10,500
6000
7000
6500
11,100
13,200
8000
10,000
10,400
12,000
13,700
18,000
20,100
19,400
24,000
3000
3500
6100
4500
7700
5500
9700
6000
7200
10,500
6000
7000
6500
11,100
13,200
8000
10,000
10,400
12,000
13700
18,000
20,100
19,400
24,000
3000
3500
6100
4500
7700
5
5
5
5
5
5
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
11
9
7
11
8
12
10
13
12
11
14
15
16
12
14
18
17
16
21
21
18
18
21
24
14
11
9
14
11
800
650
500
800
600
900
750
1000
950
850
1100
1150
1200
950
1050
1400
1350
1250
1600
1600
1400
1375
1600
1900
800
650
500
800
600
900
750
1000
950
850
1100
1150
1200
950
1050
1400
1350
1250
1600
1600
1400
1375
1600
1900
800
650
500
800
600
980
780
680
980
780
1080
880
1180
1080
980
1280
1280
1380
1080
1180
1580
1480
1380
1780
1780
1580
1580
1780
1980
980
780
680
980
780
1080
880
1180
1080
980
1280
1280
1380
1080
1180
1580
1480
1380
1780
1780
1580
1580
1780
1980
980
780
680
980
780
2200
2850
5600
3700
7100
4600
8950
5000
6250
9650
4900
5850
5300
10,150
12,150
6600
8650
9150
10,400
12,100
16,600
18,725
17,800
22,100
2200
2850
5600
3700
7100
4600
8950
5000
6250
9650
4900
5850
5300
10,150
12,150
6600
8650
9150
10,400
12,100
16,600
18,725
17,800
22,100
2200
2850
5600
3700
7100
140
140
95
140
140
140
140
140
140
140
150
150
150
140
140
150
150
150
150
150
150
150
150
150
140
140
95
140
140
140
140
140
140
140
150
150
150
140
140
150
150
150
150
150
150
150
150
150
140
140
95
140
140
1240
900
780
1240
900
1420
1060
1600
1420
1240
1780
1780
1960
1420
1600
2200
2060
1960
2520
2520
2200
2200
2520
2800
1240
900
780
1240
900
1420
1060
1600
1420
1240
1780
1780
1960
1420
1600
2200
2060
1960
2520
2520
2200
2200
2520
2800
1240
900
780
1240
900
127
(MPa)
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Pd
128
Table 1 (continued )
ID
Capacity
3
Dimensions (mm)
(MPa)
(m )
l1
l2
l3
S1
S2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
15
20
20
20
25
25
25
30
30
50
50
50
50
100
100
100
115
150
250
5
5
5
10
10
15
20
20
20
25
25
25
30
30
50
50
50
50
100
100
100
115
150
250
1800
1500
2000
1900
1700
2200
2300
2400
1900
2100
2800
2700
2500
3200
3200
2800
2750
3200
3800
1600
1300
1000
1600
1200
1800
1500
2000
1900
1700
2200
2300
2400
1900
2100
2800
2700
2500
3200
3200
2800
2750
3200
3800
5500
9700
6000
7200
10,500
6000
7000
6500
11,100
13,200
8000
10,000
10,400
12,000
13,700
18,000
20,100
19,400
24,000
3000
3500
6100
4500
7700
5500
9700
6000
7200
10,500
6000
7000
6500
11,100
13,200
8000
10,000
10,400
12,000
13,700
18,000
20,100
19,400
24,000
16
13
17
16
15
19
20
21
16
18
24
23
22
27
27
24
24
27
32
17
14
11
17
13
19
16
22
20
18
24
25
26
20
23
30
29
27
34
34
30
29
34
40
900
750
1000
950
850
1100
1150
1200
950
1050
1400
1350
1250
1600
1600
1400
1375
1600
1900
800
650
500
800
600
900
750
1000
950
850
1100
1150
1200
950
1050
1400
1350
1250
1600
1600
1400
1375
1600
1900
1080
880
1180
1080
980
1280
1280
1380
1080
1180
1580
1480
1380
1780
1780
1580
1580
1780
1980
980
780
680
980
780
1080
880
1180
1080
980
1280
1280
1380
1080
1180
1580
1480
1380
1780
1780
1580
1580
1780
1980
4600
8950
5000
6250
9650
4900
5850
5300
10,150
12,150
6600
8650
9150
10,400
12,100
16,600
18,725
17,800
22,100
2200
2850
5600
3700
7100
4600
8950
5000
6250
9650
4900
5850
5300
10,150
12,150
6600
8650
9150
10,400
12,100
16,600
18,725
17,800
22,100
140
140
140
140
140
150
150
150
140
140
150
150
150
150
150
150
150
150
150
140
140
95
140
140
140
140
140
140
140
150
150
150
140
140
150
150
150
150
150
150
150
150
150
1420
1060
1600
1420
1240
1780
1780
1960
1420
1600
2200
2060
1960
2520
2520
2200
2200
2520
2800
1240
900
780
1240
900
1420
1060
1600
1420
1240
1780
1780
1960
1420
1600
2200
2060
1960
2520
2520
2200
2200
2520
2800
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
Pd
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
129
Mb
nb
xi
x2i
i1
where xi is the distance of the i-th bolt from the neutral axis of
the plate (see Fig. 2c).
Finally, the overall action of buoyancy on the i-th bolt (Ntot,i)
may be evaluated summing the contribution of the normal net
force (Eq. (5)) and of the bending moment (Eq. (7)):
N tot;i F n;b F x;i
where Fv,b is the portion of shear force allocated to each bolt. Also
in this case, the bolts undergo a supplementary force due to the
torque caused by the vessel rotation induced by the drag force (see
130
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
Table 2
Summary of input parameters implemented in the present. study DB: derived from the database provided in Table 1, CALC: parameter calculated on the basis of input data.
Operation
ID
Item
1.1
1.2
1.3
1.4
1.5
Pd
D
L
t
l1
1.6
l3
1.7
1.8
1.9
1.10
min
max
Tst
1.11
Pst
1.12
1.13
Description/Denition
Value
0.01a-2.5
DB
DB
DB
DB
MPa
m
m
m
m
DB
1.14
1.15
ref
1.16
Vext
1.17
Vint
1.18
Wt
2.1
l2
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
S1
S2
hc
Ares
nb
fd,N
fd,S
b,adm
b,adm
Saddle height parameter (Fig. 2a) which indicates the vessel axis height
respect to the ground anchorage point
Saddle width on the vessel axis (Fig. 2c)
Transversal saddle width (Fig. 2c)
Base parameter (Fig. 2a)
Resistance area of the bolt
Number of bolts in the base plate connection
Normal design stress
Tangential design stress
Normal admissible stress
Tangential admissible stress
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
hw0
hw
hmin
hwet
vw
w
kw
tr
3.9
3.10
3.11
4.1
Fn
4.2
4.3
4.4
Fn,b
Mb
4.5
Fx,i
4.6
Ntot,i
Xi
4.7
Aext
4.8
4.9
4.10
4.11
Fv
Fv,b
Mt
yi
4.12
Fy,i
4.13
4.14
4.15
Stot,i
i
i
Units
Reference
[33]
[3133]
[3133]
[3133]
[33]
[33]
min max
0.01
0.9
300
CALC
Pa
5001500
kg/m3
[7]
CALCa
kg/m3
[54]
7800
kg/m3
[31]
1000
kg/m3
CALC
m3
CALC
m3
Given or
CALC
kg
DB
DB
DB
0.3b
157 c
410
560 c
396 c
373 c
264 c
m
m
m
mm2
MPa
MPa
MPa
MPa
04d
CALC
CALC
CALC
03.5d
1100
1.8
10500
10 3
10 2
CALC
m
m
m
m
m/s
kg/m3
CALC
y1
m
CALC
CALC
CALC
N
Nm
m
CALC
CALC
CALC
m2
CALC
CALC
CALC
N
N
Nm
m
CALC
CALC
CALC
CALC
N
MPa
MPa
[33]
[33]
[33]
[4042]
[33]
[4042]
[4042]
[4042]
[4042]
[4851]
[4851]
[36,37]
[36,37]
[52]
[52]
[55]
[55]
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
131
Table 2 (continued )
Operation
ID
Item
4.16
FC1
4.17
FC2
4.18
CFL
4.19
vw,c
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
CFL
vw,c
6.1a
6.1b
6.2
fLOC
Description/Denition
Value
2
i
b;adm
41
2 2
i
f
41
i
b;adm
i
f d;N
2
Units
Reference
[38,39]
[38,39]
CALC
CALC
m/s
CALC
CALC
CALC
CALC
CALC
CALC
CALC
CALC
m/s
CALC
d;S
Critical lling level for a given storage system and assigned storage conditions.
Repeat steps 4.1 to 4.18 in order to nd the minimum which allows satisfying
one of the two failure criteria (FC1 or FC2)
Critical velocity related to a specic vessel. Given an assigned ooding height
hw, repeat steps 4.8 to 4.18 in order to nd the minimum vw value which
allows satisfying one of the two failure criteria (FC1 or FC2)
First CFL correlation coefcient evaluated for reference substance: A K 1 Da
where K1, a are shown in Table 6
Second CFL correlation coefcient evaluated for reference substance:
B K 2 W t K 3 b where K2, K3, b are shown in Table 6:
Modied A coefcient considering a generic stored substance:
A' ref A = l v
Modied B coefcient considering a generic stored substance:
B' ref B v = l v
vw,c correlation factor: E K 4 Lc where K4, c are shown in Table 6
vw,c correlation exponent: F K 5 ln L=D K 6 where K5, K6 are shown in
Table 6
Critical lling level evaluated with correlations:CFL A'hw B'
Critical velocity evaluated with correlations: vw;c Ehw hmin F
Vessel failure probability evaluation based on critical velocity (see Fig. 7): 1
if vw Z vw,c
If vw ovw,c, vessel failure probability evaluation based on CFL (step 4.19 or 5.7,
see Fig. 7): CFL min = max min
Expected frequency of loss of containment (LOC) f LOC f
CALC
CALC
y1
10
Mt
nb
yi
11
y2i
i1
where yi is the distance of the i-th bolt from the base plate center
(see Fig. 2c).
Finally, the total shear force on the i-th bolt (Stot,i) is evaluated
summing the contribution of the shear force (Eq. (9)) and of the
torque (Eq. (11)):
Stot;i F v;b F y;i
12
41
13a
b;adm
b;adm
f d;N
2
i
f d;S
2
41
13b
where: b,adm and b,adm are respectively the normal and tangential
admissible stress; fd,N and fd,S are respectively the normal and
tangential design stress; i is the average normal stress and i is the
average shear stress on the i-th bolt derived as follows from the
loading conditions:
i N tot;i =Ares
14
i Stot;i =Ares
15
132
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
Table 3
Standard bolt connection materials classes [4042] evidencing the materials of
interest for the present application (marked with X) [33]. For symbols denition,
see Table 2.
ID Class
type
Industrial
application
fd,N
(Mpa)
fd,S
(Mpa)
b,adm
(Mpa)
b,adm
(Mpa)
A
4.6
Ba 5.6
C
6.6
D
8.8
E 10.9
X
X
X
X
240
300
360
560
700
170
212
255
396
493
160
200
240
373
467
113
141
170
264
330
hmin l2 D=2
a
Selected for the development of mechanical model and correlations (see
Table 2).
Table 4
Standard bolt dimensions range [4042] evidencing the dimensions of interest for
the present application (marked with X) [33]. For symbols denition, see Table 2.
ID Industrial
application
Bolt
Diameter
(mm)
Ares
(mm2)
A
B
C
D
Ea
F
Ga
Ha
I
12
14
16
18
20
22
24
27
30
84
115
157
192
245
303
353
459
561
X
X
X
X
X
X
o 1.2
1.22.4
42.4
18
19
a
Selected for the development of mechanical model and correlations (see
Table 2).
f 1=t r
17
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
133
% Failed tanks
16%
12%
8%
4%
0%
0%
5%
10%
15%
20%
Table 5
Reference substances, associated to the correspondent type of vessel, and ood conditions considered in the present analysis. PRES pressurized vessel; ATM atmospheric
vessel.
Parameter
ID
Reference
substances
S1
S2
S3
S4
S5
Reference
ooding
conditions
S6
W1
W2
W3
W4
Reference parameters
Description
Liquid density
l (kg/m3)
LPG
Ammonia
Gasoline or diesel fuel
Liquid aromatics or
hydrocarbons
Water solutions with
contaminant(s)
Chlorine
High depth ooding
condition
High speed ooding
conditions
Intermediate severity
ooding conditions
Low severity ooding
conditions
500
600
750
900
20.7
7.8
1.2
1.2
PRES
PRES
ATM
ATM
1100
1.2
ATM
1400
1100
24.4
PRES
0.5
2.0 10 3
1100
0.5
2.0 10 3
1100
5.0 10 3
1100
0.5
0.5
3.3 10 2
134
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
5% S1
5% S6
50% S1
50% S6
3.5
3
2.5
ID #5
V=10 m3
1.5
1
0.5
0
5% S1
5% S6
50% S1
50% S6
3.5
3
2.5
ID #20
V=100 m3
1.5
1
0.5
0
0.5
1.5
2.5
0.5
1.5
2.5
4
5% S1
5% S6
50% S1
50% S6
3.5
3
2.5
ID #53
V=10 m 3
1.5
1
0.5
0
5% S1
5% S6
50% S1
50% S6
3.5
3
2.5
ID #68
V=100 m3
1.5
1
0.5
0
0.5
1.5
2.5
0.5
1.5
2.5
Fig. 4. Example of failure plot for: (a) vessel #5; (b) vessel #20; (c) vessel #53; (d) vessel#68. Vessel data are reported in Table 1 at the correspondent ID number. Failure
plots were obtained considering two different lling levels (5 and 50%) and two different reference substances (S1,S6, see Table 5).
20
21
B K 2 W t K 3 b
22
The values calculated for the K1, K2, K3, a, and b parameters are
reported in Table 6 for each category of vessel considered. Two
sets of parameters were calculated:
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
135
23
The values of the A' and B' coefcients in Eq. (23) may be
calculated from the A and B parameters obtained using the
reference uid density ref:
A'
ref A
l v
24
B'
ref B v
l v
25
26
27
L
F K 5 ln
K6
D
28
where L is the vessel length and D is the vessel diameter. Again, for
the c, K4, K5, and K6 parameters a set of best t (SET A, dashed lines
in Fig.5) and of envelope (SET B, solid line in Fig.5) values were
calculated and are reported in Table 6.
136
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
Table 6
Parameters for CFL and vw,c evaluation applying the simplied correlations ((Eqs. (20)-28)). SET A: best t parameters; SET B: envelope correlation parameters. Pd design
pressure (MPa); ATMatmospheric pressure.
Correlation type
SET A
SET B
Vessel type
K1
K2
K3
ATM
Pd 1.5
Pd 2.0
Pd 2.5
ATM
Pd 1.5
Pd 2.0
Pd 2.5
1.331
1.287
1.290
1.256
1.331
1.347
1.341
1.355
2.163
1.144
1.305
6.068
1.882
1.197
1.365
4.512
288.6
499.2
546.0
234.0
46.8
475.8
483.6
234.0
K4
K5
K6
9.910
0.037
0.399
3.195
0.037
0.399
0.990
0.952
0.966
0.951
0.990
0.995
0.976
0.999
0.260
0.112
0.109
0.263
0.252
0.129
0.120
0.239
0.718
0.341
summarize the standard range of bolt material mechanical properties (i.e. the resistance class) and of geometrical parameters. Both
tables also report the typical values of these parameters in general
industrial practice for the anchorage of process and storage
vessels.
The sensitivity analysis was carried out applying the mechanical model, considering all the possible combinations of bolts
geometry and construction material class. Moreover, each combination was modeled considering different numbers of bolts (ranging from 4 to 10). For the sake of simplicity, as for the extended
model validation (see Section 2.4) air was assumed in the top
space of the vessels, water was assumed as the stored liquid and a
xed lling level (e.g., 10%) was considered. Fig. 6a shows the
fraction of failed vessels in the database (Table 1) varying the bolt
connection features imposing in the model the reference ooding
condition used for model validation (see Section 2.4). Since this
condition represents a threshold limit below which no failure is
expected, as mentioned in Section 2.4, no change is expected in
results, even if bolt connection parameters are changed.
Actually, the results show a signicant variation (an increase of
the fraction of failed vessels) only when low resistance connections
are considered (e.g., bolt class lower than B and Ares o200 mm2).
However, such types of connections are usually not applied in
industrial facilities for vessel anchorage [33,41]. In Fig. 6a, the
connection material type selected for the present study is highlighted by an arrow. It is evident from the gure that the results
obtained are the same when higher resistance materials with failure
fractions lower than 25% for any type of bolt with Ares 4200 mm2
are used.
In order to evaluate model uncertainties, the statistical evaluation of model predictions was carried out determining the distribution of vessels failure fraction values considering the
reference ooding conditions (see Section 2.4). All the possible
input values were implemented in the model, thus the full range
of bolt connection materials, bolts geometries and bolts number
were used. Limited input values were then considered, only
selecting the values of the parameters commonly applied in
industrial facilities according to Tables 3 and 4 [33]. In the latter
case, only 4, 8 and 10 bolts were considered for the base
connections. Fig. 6b shows the distribution of vessel failure
fractions obtained applying both the model with all possible
combinations of input parameters and ltering the input parameters according to Tables 3 and 4. The results shown in Fig. 6b
were then interpolated considering a normal distribution and
obtaining the cumulative distribution functions (CDFs) shown
respectively in Fig. 6c and d for full range and ltered input
values. As shown in Fig. 6c, if all possible input values are
considered, values of failure fraction higher than 60% are obtained,
since the weaker connections lead to the failure even in presence
of the low severity ooding. However, if only the relevant
connections types are considered, the failure fraction is lower
than 50% even in the case of weaker connections. The third
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
137
Number of observations
50
All possible input values
40
30
20
10
0
0%
Failure Fraction
CDF
1
Data
CDF
1
0.75
0.75
0.5
0.5
0.25
0.25
0
0%
20%
40%
60%
80%
100%
0
0%
Data
20%
Failure Fraction
40%
60%
80%
100%
Failure Fraction
Fig. 6. Results of the sensitivity and uncertainty analysis: (a) fraction of the tanks failed as a consequence of the validation reference ood (hw 0.5 m; vw 2 m/s)
considering different bolt connection materials (resistance class AE) and bolt sizes (Ares in mm2); (b) distribution of failure fraction values obtained applying the model with
different input types for all the vessels reported in the database (96 vessels, see Table 1); cumulated normal distribution of failure fraction values obtained applying the
model with all possible input values (c) and limiting input data (d). Density assumed for the stored uid is that of water (l 1000 kg/m3), and air is considered in the top
space of the vessel (v 1.2 kg/m3). The dashed box in panels (c) and (d) represents the third quartile (Q3) of model predictions.
29
138
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
This approach was applied to the evaluation of vertical atmospheric tanks vulnerability to ooding [8]. Either the complete
model discussed in Section 2.2 or the simplied correlations may
be applied to calculate the value of the CFL in Eq. (31).
Fig. 8a and b report parity plots obtained for the comparison of
the results obtained by the complete model and by the two sets of
simplied correlations. The data displayed in the gure were
obtained applying to all the vessels in the database (see Table 1)
a matrix of reference ood conditions (see Table 5). Data were
obtained assuming water as the stored uid (l 1000 kg/m3) and
air in the top space of the vessel (v 1.2 kg/m3). The vessels were
considered anchored (thus hw hw0).
Clearly enough, the simplied correlations allow a rapid and
straightforward assessment of vessel failure probability. On one
30
CFL min
max min
31
Evaluation of vessels
failure probability
Input data
=1
Yes
Flooding
conditions
Flood velocity
greater than
vw,c?
Vessel
geometry
No
Filling level
distribution
calculated
by Eq. (30)
0.9
CFL
0.01
UNSAFE
Operating
conditions
Operative limits
Fig. 7. Schematization of the procedure for the determination of vessels failure probability as a consequence of a given ooding scenario.
100%
Correlation SET B
Correlation SET A
100%
80%
60%
40%
W1
W2
W3
W4
20%
0%
0%
20%
40% 60%
Model
60%
40%
W1
W2
W3
W4
20%
0%
0%
80% 100%
Overpredictions
80%
20%
40% 60%
Model
80% 100%
Underpredictions
Geometric Variance
(VG)
1000
100
10
1
0.1
0.5
10
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
hand, as evident from Fig. 8a, the best t correlations (SET A),
provide results which are in good agreement with the detailed
model even if some of the prediction are not on the safe side (i.e.
failure probability is underestimated especially when values are
near to 1). On the other hand, Fig. 8b shows that the use of the
envelope correlations (SET B) allows for less accurate but always
conservative predictions of vessel failure probability.
The method proposed by Hanna et al. [51] was used to analyze
the performance of correlations in predicting the failure probability and to compare them with respect to the detailed model.
The method is based on the calculation of the geometric mean bias
(MG) and the geometric variance (VG) of the values of failure
probability predicted by the detailed model (mod) and by the
simplied correlations (corr):
"
#
mod
MG expln mod ln corr exp ln
corr
32
139
"
2 #
mod
ln
corr
33
34
As clearly appears from Eqs. (32) and (33), Eq. (34) represents
the relationship between VG and MG values in a correlation
having only a mean bias with respect to the detailed model results
(that is, a correlation for which the ratio of mod /corr is nearly
20m
River
V5-V10
V17-V19
G = GROUP OF VESSELS
G1 = V1-V4
G2 = V5-V10
G3 = V11-V16
G4 = V17-V19
V24-V26
G5 = V20-V23
G6 = V24-V26
V31
G7 = V27-V30
G8 = V31
V32-V35
V36-V39
G9 = V32-V35
G10 = V36-V39
Fig. 9. Layout of the facility analyzed in the case study. Tank data are reported in Table 7.
Fig. 10. Example of case study results: vessel failure probability (%) given the ooding conditions reported in Table 5. Panels (a) and (e): ooding W1; panels (b) and (f):
ooding W2; panels (c) and (g): ooding W3; panels (d) and (h): ooding W4. Panels (a) to (d): vessels anchored to the ground. Panels (e) to (h): vessels on a concrete base
(hc 0.3 m). Ggroup of vessels dened in Fig. 9.
2.10 10 3
NF
3.13 10 3
NF
NF
1.56 10 3
NF
2.10 10 3
NF
NF
1.68 10 3
NF
2.20 10 3
NF
NF
7.67 10 4
NF
1.68 10 3
NF
NF
NF
NF
1.16 10 3
NF
NF
6.21 10 4
NF
NF
NF
NF
1.67 10 3
1.84 10 3
2.69 10 3
9.24 10 4
3.11 10 4
2.03 10 3
3.17 10 4
1.67 10 3
1.07 10 5
3.11 10 4
1.60 10 3
8.57 10 4
2.54 10 3
6.70 10 4
8.13 10 5
1.91 10 3
1.07 10 4
1.60 10 3
NF
8.03 10 5
3.23 10 4
NF
2.43 10 3
8.66 10 4
5.51 10 5
1.86 10 3
NF
3.23 10 4
8.41 10 4
5.51 10 5
18
7.7
7.7
13.2
18
13.2
18
18
13.7
13.7
18
13
5
18
24
6
30
18
6
27
1.5
2.5
ATM
2.0
2.0
ATM
2.5
1.5
ATM
2.0
S2
S6
S5
S1
S1
S3
S1
S1
S4
S1
1.75 10 3
2.00 10 3
2.00 10 3
1.83 10 3
1.23 10 3
2.00 10 3
1.22 10 3
1.75 10 3
9.56 10 4
1.23 10 3
NF
NF
7.04 10 5
NF
NF
3.76 10 5
NF
NF
NF
NF
1.02 10 4
NF
1.33 10 4
NF
NF
4.65 10 5
NF
1.02 10 4
NF
NF
1.72 10 3
2.00 10 3
2.00 10 3
1.68 10 3
1.10 10 3
2.00 10 3
1.10 10 3
1.72 10 3
8.72 10 4
1.10 10 3
1.23 10 3
1.73 10 3
2.00 10 3
1.81 10 3
1.06 10 3
2.00 10 3
1.05 10 3
1.23 10 3
1.28 10 3
1.06 10 3
2.00 10 3
NF
1.90 10 4
NF
2.00 10 3
9.46 10 5
2.00 10 3
2.00 10 3
NF
NF
SET B
SET A
MOD
SET B
SET A
MOD
SET B
SET B
SET A
MOD
SET A
MOD
2.8
1.2
1.2
2.1
2.8
2.1
2.8
2.8
3.2
3.2
100
10
10
50
100
50
100
100
100
100
35
V1-V4
V5-V10
V11-V16
V17-V19
V20-V23
V24-V26
V-27-V30
V31
V32-V35
V36-V39
f LOC f
D
L
t
Pd
Stored
(m) (m) (mm) (MPa) Substance
ID
Capacity
(m3)
ID Vessel
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
Table 7
Case study denition and results: features of the vessels considered and expected failure frequency (y 1) due to each ooding condition dened in Table 5 considering the tanks anchored to the ground. MOD: frequency predicted
by the deterministic model; SET A and SET B refer to data predicted by the correspondent set of simplied correlations; NF: No predicted failure. Tank data were derived from the database reported in Table 1; for stored substance
features, see Table 5 at the correspondent ID.
140
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
141
Table 8
Expected failure frequency (y 1) due to each ooding condition dened in Table 5 implementing the concrete base (hc 0.3 m). MOD: frequency predicted by the
deterministic model; SET A and SET B refer to data predicted by the correspondent set of simplied correlations; NF: No predicted failure. For vessels features refer to Table 7
at the correspondent ID.
ID Vessel
V1-V4
V5-V10
V11-V16
V17-V19
V20-V23
V24-V26
V-27-V30
V31
V32-V35
V36-V39
SET A
4
8.98 10
1.66 10 3
2.00 10 3
1.37 10 3
7.48 10 4
1.74 10 3
7.26 10 4
8.98 10 4
9.99 10 4
7.48 10 4
SET B
3
1.40 10
2.00 10 3
2.00 10 3
1.26 10 3
7.83 10 4
1.77 10 3
7.85 10 4
1.40 10 3
5.85 10 4
7.83 10 4
3
1.42 10
2.00 10 3
2.00 10 3
1.39 10 3
8.96 10 4
1.82 10 3
8.89 10 4
1.42 10 3
6.70 10 4
8.96 10 4
MOD
MOD
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
SET A
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
SET B
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
4. Conclusions
A model was developed to calculate the failure probability of
horizontal cylindrical vessels as a function of ood severity.
NF
NF
1.08 10 3
NF
NF
8.01 10 4
NF
NF
1.31 10 4
NF
SET A
SET B
4
7.94 10
NF
1.22 10 3
NF
NF
8.34 10 4
NF
7.94 10 4
NF
NF
4
8.58 10
NF
1.36 10 3
NF
NF
9.54 10 4
NF
8.58 10 4
NF
NF
MOD
SET A
SET B
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
NF
Acknowledgments
The authors gratefully acknowledge nancial support received
from the EU within the 7th FP iNTeg-Risk Project (CP-IP-213345-2).
Appendix
In the following, the results of the application of the failure
model are reported. The results are summarized in charts that
allow a quick overview of the vessels response to pre-determined
ood conditions (some examples are reported in Figs. A1A4). For
each vessel, identied by the vessel ID in Table 1, the failure to a
given ood condition (identied by the imposed ood velocity vw
in m/s and effective depth hw in m) is highlighted by a red color.
On the contrary, if the model does not predict a failure, the box
is white.
142
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
Fig. A1. Failure chart for the vessels considered in the present study assuming 50% lling level and stored liquid density of 1100 kg/m3 (water solution containing toxic
contaminant).
Fig. A2. Failure chart for the atmospheric vessels considered in the present study assuming 90% lling level and stored liquid density of 1100 kg/m3 (water solution
containing toxic contaminant).
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
Fig. A3. Failure chart for the pressurized vessels considered in the present study assuming 50% lling level and stored liquid density of 600 kg/m3 (ammonia).
143
144
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
Fig. A4. Failure chart for the pressurized vessels considered in the present study assuming 90% lling level and stored liquid density of 600 kg/m3 (ammonia).
G. Landucci et al. / Reliability Engineering and System Safety 132 (2014) 125145
References
[1] Knox JC. Large increases in ood magnitude in response to modest changes in
climate. Nature 1993;361:4302.
[2] Milly PCD, Wetherald RT, Dunne KA, Delworth TL. Increasing risk of great
oods in a changing climate. Nature 2002;415:5147.
[3] Min S-K, Zhang X, Zwiers FW, Hegerl GC. Human contribution to more-intense
precipitation extremes. Nature 2011;470:37881.
[4] Seidou O, Ramsay A, Nistor I. Climate change impacts on extreme oods I:
combining imperfect deterministic simulations and non-stationary frequency
analysis. Nat Hazards 2012;61:64759.
[5] Seidou O, Ramsay A, Nistor I. Climate change impacts on extreme oods II:
improving ood future peaks simulation using non-stationary frequency
analysis. Nat Hazards 2012;60:71526.
[6] Young S, Balluz L, Malilay J. Natural and technologic hazardous material
releases during and after natural disasters: a review. Sci Total Environ
2004;322:320.
[7] Cozzani V, Campedel M, Renni E, Krausmann E. Industrial accidents triggered
by ood events: analysis of past accidents. J Hazard Mater 2010;175:5019.
[8] Landucci G, Antonioni G, Tugnoli A, Cozzani V. Release of hazardous substances in ood events: damage model for atmospheric storage tanks. Reliab
Eng Syst Saf 2012;106:20016.
[9] Rasmussen K. Natural events and accidents with hazardous materials. J Hazard
Mater 1995;40:4354.
[10] Fendler R. Floods and safety of establishments and installations containing
hazardous substancesconclusions on a research project of the German
Umweltbundesamt. Nat Hazards 2008;46:25763.
[11] Krausmann E, Renni E, Campedel M, Cozzani V. Industrial accidents triggered
by earthquakes, oods and lightning: lessons learned from a database analysis.
Nat Hazards 2011;59:285300.
[12] Cozzani V, Krausmann E, Reniers G. Other causes of escalation. In: Reniers G,
Cozzani V, editors. Domino effects in the process industries Modelling,
prevention and managing. Amsterdam: Elsevier; 2013. p. 15474.
[13] Petrova EG, Krausmann E. From natural hazards to technological disasters. Nat
Hazards Earth Syst Sci 2011;11:30635.
[14] Krausmann E, Cruz AM. Natech disasters: when natural hazards trigger
technological accidents. Nat Hazards 2008;46:139263.
[15] Renni E, Krausmann E, Cozzani V. Industrial accidents triggered by lightning.
J Hazard Mater 2010;184:428.
[16] Cruz AM, Steinberg LJ, Vetere-Arellano L. Emerging issues for natech disaster
risk management in Europe. J Risk Res 2006;9:483501.
[17] Salzano E, Iervolino I, Fabbrocino G. Seismic risk of atmospheric storage tanks
in the framework of quantitative risk analysis. J Loss Prev Process Ind
2003;16:4039.
[18] Salzano E, Garcia Agreda A, Di Carluccio A, Fabbrocino G. Risk assessment and
early warning systems for industrial facilities in seismic zones. Reliab Eng Syst
Saf 2009;94:157784.
[19] Necci A, Antonioni G, Cozzani V, Krausmann E, Borghetti A, Nucci CA. A model
for process equipment damage probability assessment due to lightning. Reliab
Eng Syst Saf 2013;115:919.
[20] Krausmann E, Cozzani V, Salzano E, Renni E. Industrial accidents triggered
by natural hazards: an emerging risk issue. Nat Hazards Earth Syst Sci
2011;11:9219.
[21] Povinec PP, Hirose K, Aoyama M. Fukushima accident radioactivity impact on
the environment. Amsterdam: Elsevier; 2013.
[22] Campedel M, Cozzani V, Garcia-Agreda A, Salzano E. Extending the quantitative assessment of industrial risks to earthquake effects. Risk Anal
2008;28:123146.
[23] Antonioni G, Bonvicini S, Spadoni G, Cozzani V. Development of a frame work
for the risk assessment of Na-Tech accidental events. Reliab Eng Syst Saf
2009;94:144250.
[24] Antonioni G, Spadoni G, Cozzani V. A methodology for the quantitative risk
assessment of major accidents triggered by seismic events. J Hazard Mater
2007;147:4859.
[25] Landucci G, Gubinelli G, Antonioni G, Cozzani V. The assessment of the
damage probability of storage tanks in domino events triggered by re. Accid
Anal Prev 2009;41:120615.
[26] Cozzani V, Gubinelli G, Antonioni G, Spadoni G, Zanelli S. The assessment of
risk caused by domino effect in quantitative area risk analysis. J Hazard Mater
2005;127:1430.
[27] Landucci G, Molag M, Reinders J, Cozzani V. Experimental and analytical
investigation of thermal coating effectiveness for 3 m3 LPG tanks engulfed by
re. J Hazard Mater 2009;161:118292.
[28] Campedel M. Analysis of major industrial accidents triggered by natural
events reported in the principal available chemical accident databases. Report
EUR 23391 EN. Ispra (I): commission of the European Communities; 2008.
145
[29] New South Wales Government. Floodplain Development Manual: the management of ood liable land. Appendix L hydraulic and hazard categorization.
Sydney: Department of Infrastructure, Planning and Natural Resources; 2005.
[30] Rijkswaterstaat. Flood risks and safety in the Netherlands (Floris), Report
DWW-2006-014. Delft: Rijkswaterstaat, Dutch Ministry of Infrastructure and
Environment; 2005.
[31] American Society of Mechanical Engineers (ASME). Boiler and pressure vessel
committee. Boiler and pressure vessel code. New York: American Society of
Mechanical Engineers; 1989 (Section VIII. Div. 1).
[32] American Petroleum Institute (API). Design and construction of large, welded,
low-pressure storage tanksAPI standard 620. Washington D.C.: American
Petroleum Institute; 2002.
[33] Sinnott RK. Coulson and Richardson's Chemical Engineering, Vol. 6. Oxford:
Chemical Engineering Design: Butterworth-Heinemann; 1999.
[34] Army US. Corps of Engineers. The great ood of 1993 post-ood report: Upper
Mississippi River and lower Missouri River basins. St. Louis: U.S.: Army Corps
of Engineers; 1993.
[35] Gruntfest E, Pollack D. Warnings, mitigation, and litigation: lessons for
research from the l993 ood. Update Water Resour 1994;95:405.
[36] Tilton JN. Fluid and particle dynamics, Section 6. Perry's Chemical Engineers'
Handbook. 7th edition. New York: McGraw-Hill; 1999.
[37] Gudmestad OT, Moe G. Hydrodynamic coefcients for calculation of hydrodynamic loads on offshore truss structures. Mar Struct 1996;9:74558.
[38] European Committee For Standardization (CEN). European Standard EN 19931-8:2005. Eurocode 3: Design of steel structures. Part 18: Design of joints.
ICS 91.010.30. Brussels: CEN; 1993.
[39] Research Council on Structural Connections (RCSC). Specication for structural
joints using high-strength bolts. Chicago: RCSC; 2009.
[40] American Society for Testing Material (ASTM). Standard specication for
carbon and alloy steel externally threaded metric fasteners. West Conshohocken: Pa: ASTM; 2004 (ASTM standard F 568M-2004).
[41] International Organization for Standardization (ISO). ISO 7411:84 hexagon
bolts for high-strength structural bolting with large width across ats (thread
lengths according to ISO 888) product grade C property classes 8.8 and
10.9. Geneva: ISO; 1984.
[42] International Organization for Standardization (ISO). ISO 7413:1984 hexagon
nuts for structural bolting, style 1, hot-dip galvanized (oversize tapped)
product grades A and B property classes 5, 6 and 8. Geneva: ISO; 1984.
[43] Kulak GL, Fisher JW, Struik JHA. Guide to design criteria for bolted and riveted
joints. 2nd ed.. Chicago: American Institute of Steel Construction, Inc; 2001.
[44] Ramachandra Rao A, Hamed KH. Flood frequency analysis. London: CRC Press;
2000.
[45] Charlton R. Fundamentals of uvial geomorphology. Abingdon, Oxon: Routledge; 2008.
[46] Bryant E. Natural Hazards. New York: Cambridge University Press; 2005.
[47] Holmes R.R. Jr. Field methods for hydrologic and environmental studies. In:
Holmes RR Jr., Terrio PJ, Harris MA, Mills PC editors. Introduction to eld
methods for hydrologic and environmental studies. Open-File Report 01-50,
Urbana: US Department of Interior U.S. Geological Survey; 2001, p. 175.
Available at http://il.water.usgs.gov/pubs/ofr01-50_chapter1.pdf.
[48] Department of Regional Development and Environment Executive Secretariat
for Economic and Social Affairs Organization of American States. Primer on
natural hazard management in integrated regional development planning.
Washington, D.C.: Dept. of regional development and environment. Executive
secretariat for economic and social affairs. Organization of American states;
1991.
[49] Pistrika AK, Jonkman SN. Damage to residential buildings due to ooding of
New Orleans after hurricane Katrina. Nat Hazards 2010;54:41334.
[50] Bates PD, Dawson RJ, Hall JW, Horritt MS, Nicholls RJ, Wicks J, et al. Simplied
two-dimensional numerical modelling of coastal ooding and example
applications. Coast Eng 2005;52:793810.
[51] Hanna SR, Strimaitis DG, Chang JC. Hazard response modeling uncertainty
(a quantitative method). Report ESL-TR-01-28. Westford, MA, USA: Sigma
Research Corporation; 1993.
[52] Italian Ministry of the Interior. Decreto Ministeriale 14/02/1997. Direttive
tecniche per l'individuazione e la perimetrazione, da parte delle regioni, delle
aree a rischio idrogeologico. Rome: Italian Ministry of the Interior; 1994.
[53] Uijt de Haag PAM, Ale BJM. Guidelines for quantitative risk assessment (Purple
Book). The Hague: Committee for the Prevention of Disasters; 1999.
[54] Liley PE, Thomson GH, Friend DG, Daubert TE, Buck E. Physical and
chemical data. (In: Perry's Chemical Engineers' Handbook). 7th ed.. New York:
McGraw-Hill; 1999.
[55] Boyce MP. Transport and storage of uids, Section 10. (In: Perry's Chemical
Engineers' Handbook). 7th ed.. New York: McGraw-Hill; 1999.