Journal of Loss Prevention in the Process Industries 28 (2014) 72e78

Contents lists available at SciVerse ScienceDirect

Journal of Loss Prevention in the Process Industries

journal homepage: www.elsevier.com/locate/jlp

Seismic vulnerability of gas and liquid buried pipelines

G. Lanzano a, E. Salzano b, *, F. Santucci de Magistris a, G. Fabbrocino a
a
University of Molise, Di.Bi.T. Department, StreGa Laboratory, Termoli, CB, Italy
b
Istituto di Ricerche sulla Combustione, CNR, Via Diocleziano 328, 80124 Napoli, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Lifelines play a crucial and essential role in human life and in economic development. The resilience of
Received 20 February 2013 those systems under extreme events as earthquakes is a primary requirement, especially when large
Received in revised form amount of toxic and flammable material are transported.
23 March 2013
In this work, the seismic vulnerability of buried gas and liquid pipelines has been analyzed, starting
Accepted 24 March 2013
from a large number of damage data to pipelines collected from post-earthquake reconnaissance reports.
Seismic fragility formulations and threshold values for the earthquake intensity with respect to the
Keywords:
release of content from different types of pipelines have been derived. The main outcome of the work is
Pipelines
Earthquake
therefore a novel seismic assessment tool which is able to cover the needs of industrial risk assessment
Seismic vulnerability procedures and land use planning requirements.
Na-Tech Ó 2013 Elsevier Ltd. All rights reserved.
Fragility

1. Introduction
A primary requirement for the industrial installations and their
fundamental components is the adequate structural safety level
with respect to natural hazard and particularly to earthquakes,
especially when large amount of toxic and flammable substances are
stored or manipulated. Among the very large number of structural
and non-structural components of the industrial plants, the pipeline
system needs special attention. Indeed, the multiple interactions
among the soil, the structure itself and the contained fluid during
seismic events is not negligible for this class of structures. Therefore,
in addition to the structural earthquake engineering, the key
contribution of geotechnical and hydraulic engineering is required
and a multidisciplinary effort is recommended.
In the past, pipelines have shown to suffer heavy damages when
loaded by seismic actions, as in the recurrent and catastrophic
earthquakes of California (San Francisco, 1906; earthquake of San
Fernando, 1971; Northridge, 1994) and Japan (Kobe, 1995). How-
ever, despite the evolution in the anti-seismic techniques and the
progress in the seismic design, relevant damages to pipelines have
been still observed during recent earthquakes occurred in Italy
(earthquake of L’Aquila, 2009; earthquake of Emilia, 2012), New
Zealand (Darfield, 2010), Chile (2010), Japan (Tohoku, 2011).
* Corresponding author. Tel.: þ39 0817621922; fax: þ39 0815704105.
E-mail addresses: salzano@irc.cnr.it, ernesto.salzano@tiscali.it (E. Salzano).
Pipelines are frequently buried for two main reasons: first, the
landfill protects the pipeline from above ground damaging events,
either natural or anthropical. Moreover, the lateral confinement
given by the surrounding soil, which increases with depth, miti-
gates the inertial seismic effects.
The present paper reports fragility curves, probit functions and
threshold values for the structural damage of buried pipelines
containing gases or liquids as function of a basic parameter of the
earthquake motion. The data and the function can be profitably
used in existing tool for the Quantitative Risk Analyses (QRA) which
includes natural-technological risks (Na-Tech issue) (Campedel,
Cozzani, Garcia-Agreda, & Salzano, 2008; Fabbrocino, Iervolino,
Orlando, & Salzano, 2005), in Land Use Planning methodologies
and more in general for the procedures activated by the public
awareness with respect to the risks related to the transportation,
transmission and distribution of hazardous substances, when
loaded by natural events (seismic Na-Tech accidents (Krausmann,
Cozzani, Salzano, & Renni, 2011; Salzano et al., 2013)).
1.1. Earthquake intensity measure
The first important aspect related to the seismic response of civil
and industrial constructions is the proper measurement of the
shaking level at the site of interest. Before quantitative seismology
was developed, the seismic intensity was estimated only on the
basis of the destructive effects on built environment and people
perceptions. That led to observational scales as the Modified

0950-4230/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jlp.2013.03.010
 G. Lanzano et al. / Journal of Loss Prevention in the Process Industries 28 (2014) 72e78 73

Table 1
Structural aspects in the seismic behavior of pipelines.

Pipelines Use Materials Joints Damage pattern

Continuous (CP) Natural gas Steel Welded joints Tension cracks
Oil Polyethylene (HDPE) Mechanical or flange joints Compression cracks
Petroleum Polyvinyl chloride (PVC) Special seismic joints Local buckling
Water Ductile iron Beam buckling
Segmented (SP) Water Asbestos cement Caulked joints Axial pull-out
Wastewater Reinforced Concrete Bell end Crushing of bell end
Polyvinyl chloride (PVC) Spigot joints. Crushing of spigot joints
Vitrified clay Circumferential failure
Cast iron Flexural failure.

Mercalli one (MMI or Macroseismic Intensity), that is still used
nowadays for its practical implications. Actually, governmental
agencies give the Macro-seismic Intensity maps for many historical
and recent earthquakes. On the basis of historical data only, many
empirical relations, developed to describe the performance of the
structures during the earthquakes, were expressed as a function of
MMI. This applies also to the seismic vulnerability of pipelines
[Eguchi, 1991; O’Rourke, Toprak, & Sano, 1998]. More recently,
instead, the deployment of seismic stations in seismic areas has led
to the more common use of basic parameters obtained from the
field measurements.
The most significant seismic instrumental parameters for
structural applications are the Peak Ground Acceleration (PGA) and
the Peak Ground Velocity (PGV), which are respectively the peak of
the horizontal acceleration and velocity time history obtained from
accelerometer measurements. Quite clearly, these parameters are
only concise descriptions of the seismic motion and do not give a
complete information of the local ground motion, in terms of fre-
quency content and signal duration, which are important param-
eter for the structural response. Despite of these limitations, due to
their simplicity, the above cited parameters are currently the most
used indexes in order to relate structural damage and seismic in-
tensity, including pipelines, and will be used in this work.
1.2. Seismic behavior of pipelines
Pipelines for the transportation of fluids are widely used for
industrial and civil purposes. These structures, having a predomi-
nant one-dimensional development, are commonly addressed as
lifelines and are often dislocated over wide areas. Few and only
general provisions are given in the current relevant codes con-
cerning the seismic behavior of these structures. In particular, the
Eurocode 8 part 4 (EN 1998-4, 2006) provides only general prin-
ciples to ensure earthquake protection. The most relevant pre-
scriptions in the context of the present paper may be summarized
as follows:
i) each structure, including pipelines, tanks and silos, must be
verified for ultimate limit state;
ii) at least one of two damage limitation states needs to be
satisfied: full integrity and minimum operating level. In the
latter case, a certain level of damage is allowed, on the basis of
a possible loss consequences of fluid or functionality;
iii) the principal safety hazard directly associated with the
pipeline rupture under a seismic event is related to explo-
sions and fires, particularly with regard to gas pipelines. The
distance of the location and the size of the population that is

Fig. 1. Summary of strong ground shaking and ground failure interaction mechanisms (adapted after O’Rourke & Liu, 1999).
74 G. Lanzano et al. / Journal of Loss Prevention in the Process Industries 28 (2014) 72e78

Table 2
Empirical correlations for Permanent Ground Displacement (d ¼ PGD).

Ground failure Seismic parameters Correlations Reference

Active fault Mw logd ¼ 4.8 þ 0.69Mw Wells & Coppersmith, 1994
Lateral spread Mw, Rep log(dh þ 0.01) ¼ 7.28 þ 1.017 Mw  0.278logR  0.026 R Bardet, Mace & Tobita, 1999
þ 0.497logY þ 0.454logS þ 0.558logT15
Landslide PGA PGA Ambraseys & Menu, 1988
logd ¼ 2:3  3:3
ac
Seismic settlement PGA PGA Takada & Tanabe, 1988
d ¼ 0:3H1 þ2
N

exposed to the impact of rupture shall be considered in software is the geographic information system-based natural haz-
establishing the level of protection; ard loss estimation software package developed by the Federal
iv) for buried pipelines, the soil/structure interaction is always Emergency Management Agency (FEMA), which is an agency of the
not negligible; United States Department of Homeland Security.
v) the hydrodynamic effects in the structures which contains The performance of the buried pipeline structure is strongly
fluid should be considered accounting all the component of related to the geotechnical effects. Based on experience and data
the motion; however, as observed for horizontal tanks, “when collected during past earthquakes, these dynamic effects can be
H/R (the ratio between filling level H and section radius R) divided in two categories [O’Rourke & Liu, 1999]: the strong ground
exceeds 1,6”, the pipeline “should be assumed to behave as if shaking (SGS), which is the transient deformation of the soil due to
it were full, i.e., with the total mass of the fluid acting solidly” wave passage (Fig. 1a), and the ground failure (GF), that is the
with the pipeline; permanent deformation of the surrounding soil due to co-seismic
vi) the use of continuous pipelines for systems which treat failure phenomena. The ground failure mechanisms appear only
flammable and pollutant material is mandatory. in specific geotechnical conditions; hence they are site dependent
and may be summarized in: active fault movement (Fig. 1b),
Damage patterns occurred in the pipelines are various and liquefaction (Fig. 1c) and earthquake-induced landslides (Fig. 1d).
largely dependent on a number of features related to the properties The Eurocode 8 provided a simplified analytical formulation to
of the material and the joint detailing. Table 1 summarises all the evaluate the maximum strains and curvatures induced in contin-
most relevant aspects from the structural perspective and shows all uous pipelines by wave passage for transient deformations as
the possible combinations of material and joints. The table also presented by Newmark (1967). The soil deformation, for each type
shows the main transportation fluids used in each structural type. and orientation of transient waves, is measured in terms of PGV.
In Table 1, two significant categories for the seismic damage are This approach is different from that employed for aboveground
reported: 1) Continuous pipelines (CP); 2) Segmented pipeline (SP). structures, as buildings, for which the structural response is directly
The hazardous, toxic and flammable, fluids must be transported related to the PGA because the importance of inertial effects in the
only in continuous pipelines, which have high strength and large seismic loadings. In Eurocode 8, however no methodology is given
deformations before structural break and consequent fluid release. for the evaluation of permanent deformations (except for some
It is worth noting that a similar approach has been already adopted indications for the design of crossing fault pipelines), in spite of the
in the context of HAZUS (FEMA, 1999), where the pipelines are fact that several references are given in the literature for this
classified as brittle (SP) and ductile (CP), depending on the seismic evaluation.
performance measured in terms of pre-failure strain levels. HAZUS An example of the available proposed correlations and related
authors are reported in Table 2. These studies proposed different
input parameters for the seismic intensity: the moment magnitude
Table 3 of the earthquake Mw, which is a unique measurement of the
Empirical pipelines SGS fragility expressions based onEq. (1). S: steel; DI: ductile
seismic size, the distance of the pipeline site from the earthquake
iron; CI: cast iron; AC: asbestos cement; C: concrete; WJ: welded joints; FJ: flange
joints; CJ: caulked joints; D: diameter.
Table 4
IM a (∙104) b Reference Empirical GF fragility expressions for pipelines based on Eq. (1). (S: steel; DI: ductile
CP SP iron; CI: cast iron; AC: asbestos cement; C: concrete; WJ: welded joints; FJ: flange
joints; CJ: caulked joints; D: diameter).
PGA e 1.698∙1012 (CI) 6.06 Isoyama & Katayama,
1982 IM A b Reference
PGV 0.3 1.0 2.25 O’Rourke & Ayala,
1993 CP SP
PGV 0.83 (S, WJ small D) 1.33 (CI, CJ, small D) 1.98 Eidinger, 1998 PGD 2.18 7.82 0.56 Honegger & Eguchi, 1992
0.25 (S, WJ, large D) 0.83 (CI, FJ, small D) PGD 3.55 (S, WJ) 23.67 (CI, CJ) 0.53 Eidinger & Avila, 1999
0.5 (DI, FJ) 0.83 (AC, FJ, small D) 16.57 (S, FJ) 16.57 (CI, FJ)
1.66 (AC, CJ, small D) 7.10 (DI, FJ) 23.67 (AC, CJ)
3.32 (AC, CJ, large D) 18.94 (AC, FJ)
1.66 (C, WJ, large D) 18.94 (C, WJ)
3.32 (C, CJ, large D) 23.67 (C, CJ)
PGV 14.5 (S, WJ, small D) 24.2 (CI, CJ, small D) 1 ALA, 2001 PGD 1.68 (S, WJ) 11.22 (CI, CJ) 0.32 ALA, 2001
3.62 (S, WJ, large D) 19.3 (CI, FJ, small D) 7.85 (S, FJ) 8.98 (CI, FJ)
16.9 (S, FJ, small D) 12 (AC, FJ, small D) 5.61 (DI, FJ) 11.22 (AC, CJ)
12 (DI, FJ, small D) 24.2 (AC, CJ, small D) 8.98 (AC, FJ)
16.9 (C, WJ, large D) 6.73 (C, WJ)
24.2 (C, CJ, large D) 11.22 (C, CJ)
19.3 (C, FJ, large D) 7.85 (C, FJ)
E 7.24∙104 0.92 O’Rourke & Deyoe, PGD e 3.21 (PVC) 0.41 Terzi, Alexoudi, & Hatzigogos,
2004 2007
 G. Lanzano et al. / Journal of Loss Prevention in the Process Industries 28 (2014) 72e78 75

Table 5 Table 7
Damage states DS for pipelines. Events used for the construction of the fragility curves.

States Hazard Patterns (structural damage) Earthquake Country Year Mw MMI PGA (g) PGV (cm/s)
DS0 Slight Investigated sections with negligible damage; San Francisco U.S. 1906 7.8 VIIIeIX 0.44e0.64 40e56
pipe buckling Long Beach U.S. 1933 6.3 VI 0.10 7
DS1 Significant Longitudinal and circumferential cracks; Kern County U.S. 1952 7.7 VIIeIX 0.17e0.57 26e112
compression joint break. Kern County U.S. 1954 5.9 VIeVII 0.12e0.29 7e28
DS2 Severe Tension cracks for continuous pipelines; joint Anchorage U.S. 1964 9.2 VII 0.20 18
loosening in the segmented pipelines. (Alaska)
San Fernando U.S. 1971 6.7 VIIIeIX 0.32e0.6 40e94
Imperial Valley U.S. 1979 6.5 IX 0.36e0.40 60e80
Michoacan Mexico 1985 8.0 VeVI 0.08e0.10 14e34
epicenter, Rep, and the PGA. Moreover, the correlations are based Whittier Narrows U.S. 1987 5.9a VII 0.28e0.36 18e24
also on topographical (as Y and S), geometrical (T15 and H1) and soil Loma Prieta U.S. 1989 6.9 VIIeVIII 0.16e0.2 40e50
Erzincan Turkey 1992 7.0 VIIIeIX 0.28e0.42 48e74
strength parameters (ac and N). Kushiro-Oki Japan 1993 7.5 VIII 0.28e0.30 36e39
Special correlations, commonly called GMPEs (Ground Motions Northridge U.S. 1994 6.7 VeX 0.08e0.76 4e164
Prediction Equations), give an estimation of some seismic site Kocaeli Turkey 1999 7.4 VIII 0.36e0.4 56e60
dependent parameter (i.e. PGA and PGV) mainly based on Mw and ChieChi Taiwan 1999 7.7 IX 0.44 96
Qinghai-Xinjiang China 2001 7.8 X 0.64 308
Rep (Douglas, 2004). The PGA and PGV are difficult to determinis-
Achaia-Ilia Greece 2008 6.5 VIII 0.34 38
tically retrieve and a direct measurement is available only for the L’Aquila Italy 2009 6.3 VIII 0.24e0.36 24e31
instrumented sites, which generally is not the same of the site of Maule Chile 2010 8.8 VIIeVIII 0.16e0.28 22e28
interest. In this way, the GMPEs were used to estimate the reference Darfield New Zealand 2010 7.1 VIeVII 0.16e0.20 16e20
seismic parameters in each site of damaged pipelines. Similarly to a
Local Magnitude.
MMI, governmental agencies released the maps for peak ground
acceleration and velocity, on the basis of deterministic measure-
ments and attenuation laws implementation, for many historical These empirical formulations for damage estimation are opti-
and recent earthquakes. These maps are commonly called Shaking mized to provide consequence assessment results mainly in terms
maps (see for instance Wald, Worden, Quitoriano, & Pankow, 2006). of repair expenditures. More interesting are the fragility functions
In the next sections the seismic vulnerability of pipelines under providing the probability of exceedance of a given damage state as a
ground failure will be expressed as a function of PGA, The choice is function of a demand parameter correlated to the seismic ground
not a limitation of this study for two main reasons: most of the motion. Therefore, in order to assess the damage probability, in the
empirical formulation for PGD estimation are directly or indirectly HAZUS approach, a Poisson probability distribution was used:
(through GMPEs) related to PGA; and the wide use of PGA as seismic
ðRR$LÞn
intensity measurement, especially from National Authorities, en- PðN ¼ nÞ ¼ eRR$L $ (2)
sures a direct and easy access to the required territorial data. n!

where P is the probability of having n pipe breaks or leaks in a pipe

2. Seismic performance in terms of repair rate
The seismic damages of the pipelines are generally described
through curves reporting a performance indicator as a function of a
seismic intensity measure. These curves are mainly derived as best
fitting of the available data coming from post-earthquake recon-
naissance. In general, the performance indicator is the repair rate,
RR, which gives the numbers of repairs after an earthquake for a
unit length of pipeline. The functional form of the curves is usually
given by:

  PGA, MMI, PGV2/PGA (Pineda-Porras & Ordaz, 2010), or even
RR n repairs=km ¼ a$IMb
where a and b are two are constants which depend on pipeline
design and characteristic as reported in Table 2 and the term IM is
an intensity indicators for the seismic action, which is mainly based
on the geotechnical aspects.
Table 6
Risk States RS for pipelines (F ¼ equivalent diameter).
segment of length L. Assuming that the pipe fails when at least one
break along its length has occurred, the probability of failure Pf is
then:
Pf ¼ 1  PðN ¼ 0Þ ¼ 1  eRR$L (3)
An interesting overview of the existing vulnerability curves for
pipelines in terms of RR was given by Tromans (2004), which
focused the attention on pipelines subjected to transient de-
formations (SGS). These formulations are mainly based on PGV as
seismic intensity indicator. However, other relations that adopt
(1) formulation based on the transient deformation of the surrounding
ground (ε), are available. Table 3 provides some empirical SGS
fragility formulations.
The empirical formulation proposed by O’Rourke and Ayala
(1993) has been also adopted by HAZUS. The expressions by
Eidinger (1998) and ALA (2001) include an additional classification
based on material, joints and tube diameter.
The fragility formulations for ground failure are fewer and
formally similar to SGS i.e., the functional relationship is given

States Hazard Patterns (loss of containment)

Table 8
Gas/Vapor/Liquefied gas Liquid
Fragility coefficients for pipelines under SGS. IM ¼ PGV expressed in cm/s.
RS0 Null No losses Limited loss
RS1 Low Very limited losses: Limited, time-distributed loss Structural aspects Class Fragility Correlation coefficient
- Toxic (F < 1 mm/m) of hazardous substance: multiple
Risk state, RS m (cm/s) b
- Flammable losses (F < 10 mm/m)
(F < 10 mm/m) CP RS1 45.22 0.39 0.996
RS2 High Non- negligible losses Large loss (e.g. entire tube surface) CP ¼RS2 71.16 0.20 0.987
or multiple losses (F > 10 mm/m) SP RS1 21.80 0.26 0.886
76 G. Lanzano et al. / Journal of Loss Prevention in the Process Industries 28 (2014) 72e78
DS0
DS1
Percent amount of data
20%
DS2
10%
0%
Fig. 2. Percent amount of data for earthquakes and DS.
again in Eq. (1). They are essentially based on Permanent Ground
Displacement, PGD, induced by the co-seismic GF phenomena
(Table 4).
The empirical formulation for Ground Failure proposed by
Honegger & Eguchi (1992) has been also adopted by HAZUS.
Despite of the various aspects accounted in the fragility formula-
tions, these existing curves are generally based on few data, except
for ALA (2001). Moreover, the repair rate RR may not be an objec-
tive parameter, because it depends on the reference length of
pipeline, which could be not uniform and comparable among the
different formulations. Finally, it is easy to recognize that either
qualitative or quantitative risk assessment of industrial facilities
and pipeline infrastructures (transportation risk) needs fragility
formulations based on different performance indicators which take
into account the structural damage and the corresponding loss of
content, particularly when the pipelines carry hazardous liquids or
gases. An approach based on such paradigm is introduced in the
following section.
3. Results
An alternative description of pipeline seismic performance is
recommended whenever attention is paid to consequence of failures
different from extension of repair interventions and related costs.
This is the case of quantitative risk analysis of industrial facilities,
which requires an estimation of released hazardous content. In such
context, an option is the construction of pipeline fragility on the
analogy with the seismic damage estimation of above ground
CP-SGS 27%
39%
SP-SGS
CP-GF
SP-GF
25%
8%
Fig. 3. Percent amount of data for each class.
industrial structures developed by Salzano, Iervolino, and Fabbrocino
(2003; 2009) and more recently by Lanzano, Salzano, Santucci de
Magistris, and Fabbrocino (2012, 2013). The method requires the
creation of a thoughtful database founded on the classification in
significant construction classes for pipelines and on the definition of
Damage State (DS) and Risk State (RS) indicators. The Damage State
refers to type of structural damage to pipeline (Table 5), whereas the
Risk States (RS) were qualified on the basis of the possible negative
effects on the external environment or population, i.e. on the
possible harmful effects derived from the release of content from the
damaged pipe (Table 6). Quite clearly, the indicators have been
distinguished on the basis of the transported fluid. In order to
correlate the structural damage to the release of containment, an
equivalent diameter F of a crack in the pipelines has been defined.
The RS levels for pipelines transporting gas, vapour and lique-
fied gas were organized in order to match the corresponding
damage states. The RS0 corresponds to DS0, in which the damage
type, even if severe, did not cause any loss of containment. The RS1
was formulated considered a very limited amount of loss, however
distinguishing between toxic and flammable substances. Finally,
the RS2 has the highest level of risk and accounts for the release of
large amount of fluid in a very short time interval. Similar risk states
RS were formulated for liquid transporting pipelines. Differently
from the previous classes, the RS0 level allows very limited loss of
liquid. The RS1 accounts limited release, but time-distributed,
100 100
(a) (b)
Limit state probability (%)
80 80
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0.2 0.4 0.6 0.8 1
PGV (cm/s) PGA (g)
Fig. 4. Fragility curves for buried pipelines, under Strong Ground Shaking (a) and
Ground Failure (b), in terms of limit state probability (%) for the RS state.,: CP
RS  RS1; >: CP RS ¼ RS2; B: SP RS  RS1; 6: SP RS ¼ RS2.
 G. Lanzano et al. / Journal of Loss Prevention in the Process Industries 28 (2014) 72e78 77

Table 9 Table 10
Fragility coefficients for pipelines under GF. IM ¼ PGA expressed in g. Probit coefficients for pipelines under SGS.

Structural Class Fragility Correlation Structural aspects Risk state, RS Probit parameters Threshold, IM0
aspects coefficient
Risk state, RS m (g) b k1 k2 (cm/s)
CP RS1 0.58 0.17 0.998 CP RS  RS1 4.12 2.41 17.05
CP ¼RS2 0.56 0.18 0.999 RS ¼ RS2 5.95 2.64 26.58
SP RS1 0.37 0.23 0.996 SP RS  RS1 0.36 1.79 5.50
SP ¼RS2 0.37 0.19 0.997

  
1 ln IM  ln m
whereas the RS2 involves the cases in which the pipe section has PðDS  DSi or RS  RSi Þ ¼ 1 þ erf pffiffiffi (4)
2 b 2
completely failed.
An extensive review of case-histories of damages for buried
where m and b are respectively the median and the shape param-
pipelines was carried out and a large database populated. The
eter of the best-fitting distribution.
dataset is composed of 335 samples, coming from edited books,
Results of the RS fragility for the four above-mentioned classes
papers and post-earthquake reports. The earthquakes analyzed for
are reported in Fig. 4. Tables 8 and 9 reports the corresponding
the construction of pipelines fragility curves, are given in Table 7. In
calculated values of the median and shape parameters. Moreover,
the table, some seismic features, in terms of range of local synthetic
the correlation coefficients between experimental data and the
parameters, are shown.
cumulative distribution has been also given; in most of the cases
In Table 8, 20 earthquakes were considered significant for
the factors are very close to unity.
pipelines damage having, in most of the cases, moment magnitude
In the case of segmented pipelines under strong ground shaking
Mw >6. The range of the intensity parameter can be very large like
the sample data were small (cfr. Fig. 3) and the correlation co-
in Northridge (1995) and San Fernando (1971) earthquakes. The
efficients are lower compared to those retrieved from other classes;
same data were plotted in a histogram graph (Fig. 2) in order to
in particular, for the cases RS ¼ RS2, a reasonable cumulative dis-
show the relative amount of data for the most important earth-
tribution was not obtained.
quakes, taking into account the DS level reported in Table 5.
Furthermore, the fragility curves obtained for the ground failure
Most of the available data came from the Californian Earthquake
are practically coincident between the RS  RS1 and RS ¼ RS2
of San Fernando 1971 (about 35%); most of the data are relative to
datasets: it means that the RS1 data are not significant for fragility
DS2 class (about 55%) and the smallest amounts are relative to DS0
estimation. This consideration, however, is in agreement to the
(about 14%).
HAZUS indication on ground failure damage, which is almost totally
Based on such a large database and on the observed behavior of
related to high damage (breaks). Also, coherently to seismic per-
buried pipelines, four possible classes of fragility curves have been
formance of segmented pipelines, the value of median m is lower
then assumed: a) continuous pipelines under strong ground shaking;
than the corresponding values of continuous pipelines. Finally, the
b) continuous pipelines under ground failure; c) segmented pipelines
value of b is higher for SGS classes, compared to GF.
under strong ground shaking; d) segmented pipelines under ground
Following the procedure given by Salzano et al. (2003; 2009), in
failure. As previously discussed, the reference intensity parameter is
order to obtain univocal threshold values both for PGV and PGA
PGV for SGS cases and PGA for GF. Accordingly, the database has been
with reference to RS states, the seismic vulnerability of the pipe-
divided into 4 classes, accounting both structural (CP or SP) and
lines has been evaluated by using the classical probit analysis. The
geotechnical aspects (SGS and GF). The relative amount of each class
probit variable Y is expressed in the Eq. (5), as a doseeresponse
is given in Fig. 3, where it is clear that most of the data are relative to
model: Y is the measure of a certain damage possibility as a func-
continuous pipelines subjected to transient deformations; instead
tion of a variable “dose” IM (Intensity Measure).
the corresponding amount of data of segmented pipelines is the
smallest, including about 25 damage cases. Y ¼ k1 þ k2 ln IM (5)
Based on this historical data and the given definition of damage
state DS and risk state RS, a uni-modal distribution of the damage Details on this classical statistical method can be found else-
states versus the intensity measure has been observed. Finally, the where in the technical literature (Finney, 1971). The resulting probit
data have been fitted by a cumulative log-normal distribution: functions are shown in Fig. 5. The calculated probit coefficients k1 e
k2 for the significant RS classes are then given in Tables 10 and 11,
where the cut-off value of the intensity measure parameters, IM0,
PGV (cm/s) PGA (g)
7 20 55 148 0.1 0.4 1.0 has also been reported. This is the value of Y providing a value of the
dose equal to 2.71, which corresponds to zero probability.
This threshold value could be considered as a strength intrinsic
parameter, significant for the pipeline structural performance and
6 6
for the soil/structure interaction during the seismic events; the
physical meaning of IM0 is defined as the limit value of the seismic
Y (-)

5 5 parameter, to above which a certain level of damage should be

considered in the Risk Analyses of the pipeline network.
4 4

Table 11
3 (a) 3 (b) Probit coefficients for pipelines under GF.

2 3 4 5 -3 -2 -1 0 Structural aspects Risk state, RS Probit parameters Threshold, IM0

LN(PGV) LN(PGA) k1 k2 (g)
CP ¼RS2 6.97 2.54 0.20
Fig. 5. Probit curves for buried pipelines under Strong Ground Shaking (a) and Ground
SP ¼RS2 7.79 2.59 0.14
Failure (b). ,: CP RS  RS1; >: CP RS ¼ RS2; B: SP RS  RS1; 6: SP RS ¼ RS2.
78 G. Lanzano et al. / Journal of Loss Prevention in the Process Industries 28 (2014) 72e78
The threshold value for SP under SGS is quite low (5.5 cm/s)
compared to corresponding value of CP (17 cm/s). In this case, the
difference between the two risk threshold levels is about 10 cm/s.
The threshold for ground failure cases increases from 0.14 g, for the
SP, to 0.2 g for continuous ductile pipelines.
This means that some attention to the behavior of existing or
new pipelines is needed in the proximity of the epicentral area, for
zones of moderate seismicity and in a quite large portion of the
territory for high hazardous seismic countries.
4. Conclusions
Industrial plants need special attention when the seismic
structural safety is under examination: therefore a possible struc-
tural damage or loss of a toxic or flammable handled material could
cause a serious risk for human life, environment and economy.
In this paper the main relevant aspects of the seismic hazard of
buried pipelines are discussed. The study of seismic behavior was
carried out through a multi-disciplinary analysis of a collected
database of damages occurred during recent documented earth-
quakes. The available damage cases were analyzed according to
geotechnical and structural relevant topics, in order to evaluate the
soil/structure interaction and estimate the pipeline response under
the seismic loadings. The final goal of the research is the con-
struction of specific fragility formulations for this class of industrial
components, accounting the multidisciplinary nature of this
research. Fragility and probit curves for pipelines are given: a novel
performance indicator was proposed, compared to the existing
fragility formulations for pipelines which are based on a global
repair rate and not on each single damage mechanism analysis.
Finally significant threshold values for seismic intensity parameters
are given, for the sake of Quantitative Risk Analyses.
Acknowledgments
This paper is dedicated to late Prof. Filippo Vinale, who has been
the head of the Science and Technologies Division (Polo delle Sci-
enze e Tecnologie) at the University of Naples Federico II and the
founder of the CIMA Center (Centro Irpino per l’Innovazione nel
Monitoraggio Ambientale).
Prof. Vinale was one of the key person in Earthquake Geotech-
nical Engineering in Italy and the pioneer of Soil Dynamics in the
country. He devoted himself with passion and care to a wide variety
of geotechnical problems, from the study of the mechanics of soils
with enhanced in situ and in laboratory testing techniques to the
full-scale analysis of embankments and dams. He was the initiator
and coordinator of profitable interactions between geotechnical
engineering and seismologist, geologist, structural engineering and
urban planners.
References
ALA, American Lifeline Alliance. (2001). Seismic fragility formulations for water
system. American Society of Civil Engineers (ASCE) and Federal Emergency
Management Agency (FEMA).
Ambraseys, N. N., & Menu, J. M. (1988). Earthquake-induced ground displacements.
Earthquake Engineering and Structural Dynamics, 16, 985e1006.
Bardet, J.-P., Mace, N., & Tobita, T. (May 4, 1999). Liquefaction-induced ground
deformation and failure. Report to PEER/PG&E, Task 4A e Phase 1.
Campedel, M., Cozzani, V., Garcia-Agreda, A., & Salzano, E. (2008). Extending the
quantitative assessment of industrial risks to earthquake effects. Risk Analysis,
28(5), 1231e1246.
Douglas, J. (2004). Ground motion estimation equations 1964e2003. Reissue of ESEE
Report No. 01e1: ‘A comprehensive worldwide summary of strong-motion atten-
uation relationships for peak ground acceleration and spectral ordinates (1969 to
2000)’ with corrections and additions. Research report 04-001-SM. London, UK:
Imperial College.
Eguchi, R. T. (1991). Early post-earthquake damage detection for underground lifelines.
Final Report to the National Science Foundation. Los Angeles, CA: Dames and
Moore PC.
Eidinger, J. (1998). Lifelines, water distribution systems in the Loma Prieta, California,
earthquake of October 17, 1989. Performance of the built environment e Lifelines.
US Geological Survey Professional Paper 1552-A (pp. 63e80).
Eidinger, J., & Avila, E. (1999). Guidelines for the seismic upgrade of water trans-
mission facilities. TCLEE Monograph, 15.
EN 1998-4. (2006). Eurocode 8: Design of structures for earthquake resistance e Part
4: Silos, tanks and pipelines. CEN European Committee for Standardization.
Fabbrocino, G., Iervolino, I., Orlando, F., & Salzano, E. (2005). Quantitative risk analysis
of oil storage facilities in seismic areas. Journal of Hazardous Materials, 23, 61e69.
FEMA. (1999). Earthquake loss estimation methodology. HAZUS-MH. Technical
manual, retrieved on. www.fema.gov/hazus.
Finney, D. J. (1971). Probit analysis (3rd ed.). Cambridge, UK: Cambridge University Press.
Honegger, D. G., & Eguchi, R. T. (October 1992). Determination of the relative vul-
nerabilities to seismic damage for San Diego County Water Authority (SDWCA).
Water transmission pipelines.
Isoyama, R., & Katayama, T. (1982)Reliability evaluation of water supply systems during
earthquake, report of the Institute of Industrial Science, Vol. 30, University of Tokyo.
Krausmann, E., Cozzani, V., Salzano, E., & Renni, E. (2011). Industrial accidents
triggered by natural hazards: an emerging risk issue. Natural Hazard Earth
System Sciences, 11, 921e929.
Lanzano, G., Salzano, E., Santucci de Magistris, F., & Fabbrocino, G. (2012). An
observational analysis of seismic vulnerability of industrial pipelines. Chemical
Engineering Transactions, 26, 567e572.
Lanzano, G., Salzano, E., Santucci de Magistris, F., & Fabbrocino, G. (2013). Seismic
vulnerability of natural gas pipelines. Reliability Engineering & Systems Safety.
in press.
Newmark, N. M. (1967). Problems in wave propagation in soil and rocks. In Proc. Int.
Symp. On wave propagation and dynamic properties of earth materials (pp. 7e26).
University of New Mexico Press.
O’Rourke, M. J., & Ayala, G. (1993). Pipeline damage due to wave propagation.
Journal of Geotechnical Engineering, 119, 123e134.
O’Rourke, M. J., & Deyoe, E. (2004). Seismic damage to segment buried pipe.
Earthquake Spectra, 20-4, 1167e1183.
O’Rourke, M. J., & Liu, X. (1999). Response of buried pipelines subjected to earthquake
effects. MCEER Monograph No.3. Buffalo, USA: University of New York.
O’Rourke, T. D., Toprak, S., & Sano, Y. (1998). Factors affecting water supply damage
caused by the Northridge Earthquake. In Proceedings 6th US national conference
on earthquake engineering. Seattle, Washington (pp. 1e12).
Pineda-Porras, O., & Ordaz, M. (2010). Seismic fragility formulation for segmented
buried pipeline systems including the impact of differential ground subsidence.
Journal of Pipeline System Engineering Practice, 1, 141e146.
Salzano, E., Basco, A., Busini, V., Cozzani, V., Renni, E., & Rota, R. (2013). Public
awareness promoting new or emerging risk: industrial accidents triggered by
natural hazards. Journal of Risk Research. in press.
Salzano, E., Garcia Agreda, A., Di Carluccio, A., & Fabbrocino, G. (2009). Risk
assessment and early warning systems for industrial facilities in seismic zones.
Reliability Engineering & Systems Safety, 94, 1577e1584.
Salzano, E., Iervolino, I., & Fabbrocino, G. (2003). Seismic risk of atmospheric storage
tanks in the frame work of quantitative risk analysis. Journal of Loss Prevention
in the Process Industry, 16, 403e409.
Takada, S., & Tanabe, K. (1988). Estimation of earthquake induced settlement for
lifeline engineering. InProc. of the 9th World Conference on Earthquake Engi-
neering, August, Vol. VII, (pp. 109e114).
Terzi, V., Alexoudi, M. N., & Hatzigogos, T. N. (2007). Numerical Assessment of
Damage State of segmented pipelines due to permanent ground deformation. In
Proc. of 10th international conference on applications of statistics and probability
in civil engineering, Tokyo, Japan, paper no. 202.
Tromans, I. (2004). Behavior of buried water supply pipelines in earthquake zones.
PhD Thesis. Imperial College of Science, Technology and Medicine, University of
London.
Wald, D. J., Worden, B. C., Quitoriano, V., & Pankow, K. L. (2006). ShakeMapÒ manual.
Technical manual, users guide, and software guide. Advanced national seismic
system.
Wells, D. L., & Coppersmith, K. J. (1994). New empirical relationships among
magnitude, rupture length, rupture width, rupture area and surface displace-
ment. Bulletin of Seismological Society of America, 84(4), 974e1002.