HazMat Transportation Risk Assessmen
HazMat Transportation Risk Assessmen
HazMat Transportation Risk Assessmen
a r t i c l e i n f o a b s t r a c t
Article history: The Seveso accident triggered a virtuous process towards the development of methods, models and tools
Received 8 May 2016 for safety and risk assessment and management. Among the more relevant results of such process was
Received in revised form the stemming of methods and tools addressing the transportation of hazardous substances in the
30 July 2016
framework of a holistic approach to the control of major accident hazards related to dangerous sub-
Accepted 14 August 2016
stances. The present study aims at the analysis of reference procedures and tools available for the
Available online 18 August 2016
analysis of the risk in the transportation of dangerous substances in the light of the Viareggio accident.
The Viareggio accident represents a paradigmatic event involving the transportation of dangerous
Keywords:
Hazardous materials transportation
substances. The accident, that took place in Italy in 2009, was analyzed in the perspective of current
Quantitative risk assessment approaches to the analysis of risk in the transportation of hazardous materials. The results pointed out
Major accident hazards that the Viareggio scenario, although of particular severity, is comprised within those accounted in
Accident data analysis quantitative risk analysis.
Transportation risk analysis © 2016 Elsevier Ltd. All rights reserved.
1. Introduction et al., 2003, 2000). Thus, in parallel with quantitative risk assess-
ment of fixed installations, several methods and tools were devel-
Process safety and control of major accident hazards related to oped for the quantitative assessment of risk due to the
dangerous substance made tremendous progresses since the 70s. A transportation of hazardous substances. Different methodological
few accidents played an important role in calling for the advance- approaches were defined for transportation risk analysis (TRA) and
ment of methods, tools and regulations for safety assessment and several countries adopted specific regulations to manage and
safety management in process plants. The Seveso accident is one of control the risk due to HazMat transportation. As evident from the
such milestone events, and the wide body of regulations issued in comprehensive literature reviews available, the approaches more
the European Union to control major accident hazards are still frequently applied take into account the combined evaluation of
named “Seveso” Directives (European Commission, 2012, 2003, the expected accident frequency and the consequent impact on
1997, 1982). Although the transportation of hazardous substances population of accidents during the HazMat transportation (ACDS,
was since the beginning excluded from the application of Seveso 1991; CCPS - Center of Chemical Process Safety, 1995; CFR Code of
Directives, the need for a holistic approach to risk assessment and Federal Regulations, 2015; CGSB - Canadian General Standards
management was soon evident, and a number of pilot studies Board, 1997; Erkut et al., 2007; Gheorghe et al., 2005; Lees, 1996;
demonstrating the importance of accounting for the transport of OPSI, 2009; Purdy, 1993; Uijt de Haag and Ale, 1999; VROM, 1996).
hazardous substances to and from Seveso sites in industrial areas Several site-specific studies in critical areas are available in the
and chemical clusters were carried out (Egidi et al., 1995; HSE- literature to support the implementation of the proposed meth-
Health Safety Executive, 1981, 1978; RP Authority, 1982; Spadoni odologies. The risk study for Canvey Island (HSE-Health Safety
Executive, 1981, 1978) focused on the risk evaluation associated to
fixed and harbor activities, in order to establish the acceptability of
* Corresponding author. the existing industrial installations and to identify risk reduction
E-mail address: valerio.cozzani@unibo.it (V. Cozzani).
http://dx.doi.org/10.1016/j.jlp.2016.08.009
0950-4230/© 2016 Elsevier Ltd. All rights reserved.
G. Landucci et al. / Journal of Loss Prevention in the Process Industries 49 (2017) 36e46 37
policies. The pilot study on Rijnmond area (RP Authority, 1982) accident.
allowed consolidating the adoption of risk assessment methods
and models to the control of industrial hazards. An integrated area 2. Methods
risk assessment was carried out among the ARIPAR project, focused
on the Ravenna industrial area in Italy (Egidi et al., 1995), imple- 2.1. Past accident analysis
menting the analysis of risk induced by road, rail, ship and pipeline
transportation of hazardous materials. Past accident analysis is a widely used tool for preliminary
More recently, Guo and Verma (2010) investigated the impact of hazard identification in chemical facilities (CCPS - Center of
different truck capacities on transport risk, supporting the man- Chemical Process Safety, 1992). In the present study, it was used
agement of flammables and explosives transportation in Illinois, to demonstrate its effectiveness in analyzing scenarios associated
USA. TNO (Boot, 2013) developed a risk based methodology to to the transportation of hazardous materials (HazMat) and to un-
compare the different transportation modes and applied it to derstand if the reference case of Viareggio accident was “one of the
ammonia and LPG (liquefied petroleum gas) transportation in the kind” or a common scenario in dangerous goods transportation.
Dutch context. The MHIDAS (AEA Technology, 2001), FACTS (Unified Industrial &
Simplified quantitative and semi-quantitative assessment Harbour Fire Department Rotterdam-Rozenburg, 2016), and ARIA
methods based on the combined evaluation of expected accident (Ministe re de l'Ecologie du De
veloppement durable et de l'Energie,
frequency and the consequent impact on population were also 2016) databases were used to collect data on accidents that took
proposed. Erkut and coworkers (Erkut and Verter, 1998; Erkut et al., place on a 20 years period (1998e2008). No substantial change
2007) presented a simplified risk model which adopted a conse- took place in this period in RID (Regulations concerning the inter-
quence assessment based on danger circles and defining unit road national carriage of dangerous goods by rail) (European
segment risk and the edge risk. Bubbico et al. (2004a) showed an Commission, 2006), that apply to the transportation of LPG by
approach based on the definition of a limited set of route- rail in European countries. A total of 362 accidents were analyzed.
independent information (typical weather conditions, release sce-
nario probabilities, impact of predetermined accidents, etc.) in or- 2.2. Reference assessment of risk in LPG railway transportation
der to rapidly perform TRA. The approach was implemented on a
GIS-based tool (Bubbico et al., 2004b) and applied for the risk Several well defined procedures are available for the assessment
management of road and rail transport of hazardous materials in of risk due to the rail HazMat transportation (ACDS, 1991; CCPS -
Sicily (Bubbico et al., 2006). Fabiano et al. (2005) showed a site- Center of Chemical Process Safety, 1995; Lees, 1996; Uijt de Haag
oriented framework for risk assessment proposing a theoretical and Ale, 1999), and a significant number of methodological
approach for emergency planning and optimization based on graph studies (Assadipour et al., 2015; Erkut and Ingolfsson, 2005; Verma,
theory application. 2011, 2009) and applications (e.g., see Bubbico et al. (2006), Cozzani
The security related aspects of hazardous materials trans- et al. (2007), Milazzo et al. (2010, 2002)) are available. A detailed
portation, e.g. considering accidents triggered by malevolent acts of discussion of the procedure for the risk assessment of LPG rail
interference, were accounted for with a dedicated security transportation is out of the scope of the present study. However, a
vulnerability assessment methodology described in Reniers and few key-elements will be briefly recalled in the followings, in order
Dullaert (2013), Reniers and Zamparini (2012). Therefore, the to understand which accident scenarios are usually analyzed in the
problem of HazMat transportation risk analysis is well known. Both risk assessment of LPG transportation.
consolidated and innovative methodologies are available and Several TRA studies only focused on the consequence analysis of
applied. the accidental scenarios which may affect the surrounding popu-
However, even if the application of such methods and tools lation during HazMat shipment (Erkut and Ingolfsson, 2005; Erkut
allowed the assessment and, in many cases, the control and miti- et al., 2007), in order to perform evacuation and emergency
gation of risk due to HazMat transportation, public concern on the response capability assessment (Verma and Verter, 2007).
safety of HazMat transportation raised worldwide after some se- On the other hand, other authors (Saccomanno and Shortreed,
vere accidents took place in Lac-Me gantic, Canada (U.S. 1993; Schweitzer, 2006) pointed out that variation in population
Department of Transportation, 2013), Wetteren, Belgium (De density around the transport network may be neglected or, in some
Smedt et al., 2014), and Viareggio, Italy (Landucci et al., 2011). cases, the expected damage distance associated to the worst case
These events generated important issues on the adequacy of safety scenarios is small. In such studies (e.g. see Saccomanno and Chan
standards for HazMat transportation and, more in general, on the (1985)), the TRA is only performed on a probabilistic basis, in or-
adequacy of the current approaches for HazMat transportation risk der to determine the factors which might minimize the accident
analysis. More recently, the use of quantitative risk assessment was occurrence frequency.
criticized due to the inadequacy of the methodology to answer However, the approach more frequently applied to TRA is based
social expectations concerning the protection of the population and on the combined estimation of frequencies and consequences (Alp,
the safety of railway transportation. The results were deemed to be 1995; Batta and Chiu, 1988; Spadoni et al., 1995). Table 1 briefly
of limited value since focusing only on casualties and affected by outlines the features and the key elements of several methodolo-
uncertainty, as well as more difficult to communicate with respect gies which apply the mentioned risk approach with the corre-
to all-hazard semi-quantitative methodologies (Houdijk, 2016). spondent reference.
In this framework, the present study proposes an assessment of In this study, in order to systematically compare the potential-
the different methodologies for TRA in the perspective of the ities of the current methodologies, the contribution of different LOC
Viareggio accident, in order to verify the adequacy of the current (loss of containment) scenarios and final outcomes to risk obtained
tools and procedures. In particular, attention was specifically for LPG transportation in a densely populated area was analyzed. In
focused on consequence analysis in order to evidence possible particular, two standard data sets were used as an input to the
limitations related to the use of conventional simplified point- analysis: i) “Purple Book” data (Uijt de Haag and Ale, 1999); and ii)
source models. Possible prevention and mitigation measures to data reported in (Bonvicini et al., 2002; Cozzani et al., 2004a,
control the risk deriving from HazMat transport accidents are also 2004b) used in a previous TRA study carried out on the Italian
discussed taking into account the event sequence of the Viareggio railway network. F-N curves (see definition in Table 1) were
38 G. Landucci et al. / Journal of Loss Prevention in the Process Industries 49 (2017) 36e46
Table 1
Alternative tools and procedures to perform transportation risk analysis of hazardous substances.
1) Characterization of the a) Determination of transport path, estimation of HazMat transportation fluxes (Cozzani et al., 2007; Leonelli et al.,
transportation path b) Evaluation of population distribution around the transportation path 1999; Uijt de Haag and Ale, 1999)
c) Determination of reference meteorological conditions of the considered site
2) Loss of containment (LOC) Tanker puncture due to collision or derailment, leading to an instantaneous release or a large (Uijt de Haag and Ale, 1999)
events considered in the section release
analysis Failure of components or misoperation of equipment e continuous release with limited (ACDS, 1991; Lees, 1996; Uijt de Haag
release section and Ale, 1999)
Nominal equipment leak (Purdy, 1993)
3) LOC Frequency evaluation Evaluation of frequency by the use of past accidents data analysis (Erkut et al., 2007; Lees, 1996)
Use of detailed logic tools (fault tree analysis) (Lees, 1996; Rhyne, 1994)
Analytical and simulation modeling (Gheorghe et al., 2005)
Formal subjective estimation (CCPS - Center for Chemical Process
Safety, 2008)
5) Consequence evaluation Impact area based on pre-determined geometrical shapes (see [a]): [a] (Erkut et al., 2007)
- band of fixed width around each route segment [b] (U.S. Department of
- danger circle, with a substance-dependent radius cantered at the incident location Transportation, 2012)
- rectangle around the route segment
- ellipse shape based on the Gaussian plume model
Damage distance is based on the evacuation distance (i.e., the initial isolation zone) when a
hazmat incident occurs, i.e., 0.8 km for flammable HazMat and 1.6 km for flammable and
explosive HazMat [b]
Consequence analysis identifying final outcomes with event tree analysis and calculation of (ACDS, 1991; Lees, 1996; Uijt de Haag
physical effects map with lumped parameters models and Ale, 1999)
6) Impact on population Impact area based on threshold of physical effect (Brown and Dunn, 2007; Gheorghe
et al., 2005)
Probit models to determine the vulnerability zones around the railway (Uijt de Haag and Ale, 1999; Van Den
Bosh, 1989)
7) Risk recomposition Individual risk profile around the transport path (ACDS, 1991; Uijt de Haag and Ale,
1999)
Societal risk evaluating F-N plots, reporting the cumulative expected frequency (F) of accidents (ACDS, 1991; Uijt de Haag and Ale,
having a given number of fatalities (N) 1999)
Expectation value EV defined as: (ACDS, 1991; Carter and Hirst, 2000;
PNmax Leonelli et al., 1999)
EV ¼ N¼1 FðNÞ
Nmax ¼ maximum number of fatalities in the F-N plot
Risk Integral RI defined as: (ACDS, 1991; Carter and Hirst, 2000;
P max
RI ¼ N N¼1 FðNÞ$N
Leonelli et al., 1999)
calculated in order to evaluate the societal risk for a representative Calculations were carried out using conservative conventional
case-study. The case-study was selected to give an example of the assumptions: LPG was considered composed only of propane. Sta-
typical risk figures that may be caused by the transportation of LPG bility class F and a wind velocity of 2 m/s were assumed, and a
by railway across a densely populated area in Italy, as that of horizontal release was considered for jet fires. As shown in the
Viareggio. The case-study is described in the Supplementary table, flash fire is among the scenarios causing the higher damage
Material. distances from the release point, up to about 300 m from the
release position. A specific discussion of the damage distances
compared with the actual extension of damages reported in the
2.3. Reference damage distances Viareggio accident will be carried out in Section 4.3.
Threshold values reported in Table 2, although adopted in the
The introduction of standard LOC categories in the risk analysis legislation of some European countries falling under the obligation
of LPG transportation allows for the calculation of the corre- of Seveso directive (European Commission, 2012), are uncertain
sponding damage distances adopting a threshold based approach. and somehow controversial from a scientific and technical point of
Damage distances shown in Table 2 were calculated for each final view. An alternative approach to understand the extension of zones
outcome assuming the LOC categories selected for the case study that should be considered as potentially affected by the
(see Supplementary Material).
Table 2
Damage distances calculated for the release of propane using the standard approach proposed for HazMat risk assessment in a previous Italian study (Bonvicini et al., 2002;
Cozzani et al., 2004a, 2004b). A horizontal jet-fire and cloud ignition at release position were considered.
Release Category Final outcome Threshold for start of lethality (Ministero dei lavori Pubblici, 2001) Damage distance (m)
2
R1 (30 mm) Jet fire 7 kW/m 50
Flash fire 0.5 LFL 50
VCE 0.14 bar 25
R2 (80 mm) Jet fire 7 kW/m2 110
Flash fire 1/2 LFL 90
VCE 0.14 bar 90
R3 (Instantaneous) Fireball 350 kJ/m2 250
Flash Fire 0.5 LFL 280
VCE 0.14 bar 200
G. Landucci et al. / Journal of Loss Prevention in the Process Industries 49 (2017) 36e46 39
consequences of an LPG release may be based on the calculation of liquids (RID class 3) and liquefied gases (including LPG, RID class 2).
a composite “lethality map”. For each of the final outcomes re- Table 4 shows that the yearly quantities of liquefied gases trans-
ported in Table 2, a vulnerability (death probability) or lethality ported by rail in Europe and in North America are comparable.
map may be calculated on the basis of human vulnerability models, However, North America is characterized by average shipping dis-
widely used in risk assessment procedures (Lees, 1996; Van Den tances that are one order of magnitude higher than in Europe.
Bosh, 1989). In Section 4.3, the predicted lethality levels evaluated Table 5 shows that LPG and other liquefied flammable gases are
for the case-study analyzed in the Supplementary Material are among the substance categories more frequently involved in Haz-
compared with the actual impact area determined for the Viareggio Mat railway accidents. Even if data from railway boards indicate
accident on the basis of detected damages. that incidents at marshalling yards or in stations are more frequent
than en-route accidents (Ruffin et al., 2009), Table 6 shows that the
3. The Viareggio accident events reported in the MHIDAS database for railway transportation
of HazMat can mostly be associated with en-route accidents. It
A detailed description of the Viareggio accident is reported in should be recalled that the reporting criteria of MHIDAS database
the literature (Brambilla and Manca, 2010; Manca and Brambilla, may lead to underreporting of events not involving the release of
2010; Manca et al., 2010; Tugnoli et al., 2013). Hereby the key as- hazardous substances (AEA Technology, 2001). Possibly, the high
pects of the accident are briefly summarized. speed of trains involved in en-route accidents, typically in the range
On Monday, June 29th, 2009 train 50325 was delivering LPG 90e120 km/h both in North America (Kawprasert and Barkan,
from the oil refinery of San Martino di Trecate, near Milan, to the 2010) and Europe (European Commission, 2006), may more easily
LPG storage of Gricignano, near Naples. The train was composed of result in tank puncturing and release with respect to passage in low
14 tank wagons owned by GATX Rail, each having a nominal ca- speed tracks (speed lower than 40 km/h (Uijt de Haag and Ale,
pacity of 46.7 t (approximately 110,000 L). Around 11:45 p.m. the 1999), such as in the case of marshalling yards).
first tank wagon of the train derailed about 1 km before the station The longer shipping distances may be the cause of the different
of Viareggio. The closed circuit cameras of Viareggio station accident frequency (events/year) among North America and
recorded the full-speed passage of the derailed train at 11:48 p.m. Europe. The conclusion seems to be confirmed by the higher
(Landucci et al., 2011). The derailed tankcar overturned while occurrence of en-route accidents in North America (see Table 6),
passing through the shunts after the railway station and conse- which are almost four times more frequent than in Europe. How-
quentially also the following four tank cars derailed and over- ever, the average shipping length in North America is also about
turned. The train engine did not derail and stopped few meters four times higher than Europe (data elaborated from Ruffin et al.
ahead of the first car. (2009), U.S. Department of Transportation and U.S. Department of
After the derailment, an intense loss of containment took place Commerce (2007)), hence the approximate accident rate per unit
from the first tank wagon. The entire inventory of the ruptured length appears to be homogenous among the two considered
tankcar was released from a breach (approximate equivalent frameworks.
diameter of 150 mm (Pontiggia et al., 2011)) caused by the impact Table 7 reports the data collected from different accident data-
following the derailment. No loss of containment occurred from the bases on railway accident scenarios involving liquefied flammable
other 13 tank vessels. gases (84% of which actually involving LPG).
A delayed ignition occurred few minutes after the release. The The table evidences that flash-fire was already recorded as a
formation of a gas cloud is reported by several witnesses, that possible final outcome in 3% of the cases. Moreover, if also VCEs are
remember a cloud of cold and white mist propagating around the considered, delayed ignition of unconfined vapor clouds occurred
area where the train derailment took place (Landucci et al., 2011). It in 7% of the cases (14 events on a 50 years period).
is still uncertain which was the ignition point and if a single or more Therefore, en-route transit accidents emerge as credible events
than one ignition sources were present. A flash fire resulting in affecting the transportation of hazardous materials, likely to lead to
severe damages, also affecting the houses and vehicles close to the the release of transported substances. Flash fire is a previously
railway, took place. A total of 32 fatalities were caused by the fire. experienced final outcome of LPG releases during railway trans-
portation. Thus, the Viareggio scenario should be among those
4. Results and discussion expected and considered in the analysis of risk due to LPG railway
transportation.
4.1. Past accident data analysis
4.2. Societal risk caused by LPG rail transportation
Table 3 shows that the two more important categories of haz-
ardous materials transported by railway in Europe are flammable Fig. 1 reports the F/N curves calculated from available records for
Table 3
Quantities (109 kg km railway) of hazardous materials falling under RID regulations transported in the 27 countries of European Union (Ruffin et al., 2009).
Class 1: Explosive substances and articles 1743 2.7 833 1.3 1096 1.7
Class 2: Gases (compressed, liquefied, dissolved under pressure) 8625 13.2 8073 12.8 8064 12.8
Class 3: Flammable liquids 35,127 53.6 36,767 58.3 36,313 57.6
Class 4: Flammable solids þ Substances liable to spontaneous combustion þ Substance emitting flammable gases with 2354 3.6 2749 4.4 3541 5.6
water
Class 5: Oxidising substances þ Organic peroxides 5156 7.9 2530 4.0 3402 5.4
Class 6: Toxic substances þ Substances liable to cause infections 1527 2.3 1748 2.8 1638 2.6
Class 7: Radioactive material 832 1.3 126 0.2 315 0.5
Class 8: Corrosive substances 3362 5.1 4948 7.8 4347 6.9
Class 9: Miscellaneous dangerous substances and articles 6751 10.3 5267 8.4 4284 6.8
All classes 65,477 100.0 63,041 100.0 63,000 100.0
40 G. Landucci et al. / Journal of Loss Prevention in the Process Industries 49 (2017) 36e46
Table 4
Quantities of Class 2 goods (Gasses: compressed, liquefied, dissolved under pressure) transported by rail in year 2007 (elaborated from Canada (2011),
n General de Transporte Ferroviario y Multimodal (2007), Eurostat (2012), U.S. Department of Transportation and U.S. Department of
Direccio
Commerce (2007)).
Table 5
Substances involved in accidents during HazMat transportation by rail. Data refer to 362 accidents reported in MHIDAS database for the period 1988e2000.
Table 6
Location of accidents involving HazMat transportation by rail. Data refer to 362
accidents reported in MHIDAS database for the period 1988e2000.
Table 7
Final outcome scenarios of accidents involving flammable pressurized liquefied gas
transportation by rail. Data refer to 201 accidents reported in MHIDAS, ARIA, FACTS,
databases for the period 1960e2010.
Scenario Events %
Fig. 1. Societal risk (F/N curves) derived from the analysis of past accidents involving
BLEVE (of which with fireball) 49 (13) 24% (6.5%) rail and road transportation of flammable pressurized liquefied gases. Data refer to
Flash fire 6 3% accidents reported in MHIDAS, ARIA and FACTS databases for the period 1950e2010.
Vapor Cloud Explosion (VCE) 8 4%
Partially Confined-Explosion 1 0.5%
Jet Fire 23 11.5%
Fire and/or explosion, detail non specified 22 11% figure are specific of the case-study considered, they allow drawing
Dispersion 30 15% some general conclusions concerning the influence of the different
No release 62 31% scenarios on societal risk. In particular, Fig. 2a shows the F-N curves,
while Fig. 2b reports the contribution of each scenario to the overall
value of societal risk.
accidents in transportation of flammable pressurized liquefied As shown in the figure, the more severe scenario expected in the
gases. case of LPG transportation by rail is the fireball that may follow a
The low number of events required to consider both road and BLEVE (Birk and Cunningham, 1996; Maillette and Birk, 1996;
rail transportation accidents in the plot. Though recorded flash fire Paltrinieri et al., 2009). However, flash fires and/or VCEs following
scenarios clearly show a tendency toward multiple fatalities, the either a continuous or an instantaneous release give a relevant
Viareggio accident resulted in a considerably higher death toll. contribution to the societal risk, being responsible of final out-
However, carefully considering the curves in Fig. 1, the Viareggio comes having relevant consequences (N ranges from 20 to 200 in
accident seems to be in line with the F-N curve of flash fire sce- the case-study examined). Moreover, the F-N societal risk curve
narios as well as with the statistical profile of the other severe but shows that the overall expected frequencies of these rather severe
extremely rare accidents occurred in the transportation of flam- scenarios are about two orders of magnitude higher than those of
mable liquefied gases (curves for BLEVE e Boiling liquid expanding the fireball. As a matter of fact, Fig. 2b shows that the contribution
vapor explosion and VCE e vapor cloud explosion). of flash fires and VCEs ranges up to 62% of the total PLL calculated
In order to obtain more specific data on the risk profile due to for the case-study.
LPG transportation, Fig. 2 summarizes the results of the analysis of Therefore, both the results obtained by implementing the his-
the case-study discussed in Section 2 and presented in the torical records and by the calculations for the case study evidence
Supplementary Material. Although the results reported in the that flash-fires are among the scenarios that, although having an
G. Landucci et al. / Journal of Loss Prevention in the Process Industries 49 (2017) 36e46 41
Fig. 2. Societal risk calculated for the case-study described in the Supplementary
Information: a) F-N plot (F: overall frequency, 1/y; N: expected number of fatalities);
b) Contribution to Potential Life Loss calculated from the F-N plot for the final out-
comes considered (events/106 years and % of total value). FF ¼ flash fire; VCE ¼ vapor
cloud explosion.
radiation and partial flame engulfment were experienced. The zone Fig. 4b shows a comparison of the lethality curves, introduced in
is evidenced in yellow in Fig. 3 and extends for about 430 m in the Section 2.3, with the damage map of the Viareggio accident. As
north-south direction and for about 190 m in the east-west direc- shown in the figure, the red zone is almost entirely comprised
tion. The red and yellow areas may be considered as those where between the 10% and the 1% death probability curves. This on one
the actual flash-fire took place (about 0.1 km2). Finally, a zone in hand confirms that the impact area of the Viareggio accident falls
which only radiation damages could be documented (green zone in inside the area where fatalities were expected by the current risk
Fig. 3) is slightly more extended than the yellow zone, confirming assessment methods. On the other hand, it should be recalled that
that in flash-fires a limited damage is expected outside flame the lethality curves shown in Fig. 4b are average values obtained
boundaries. considering the conditional probabilities of all possible final out-
As previously mentioned, in order to test the credibility of the comes. The comparison of the average death probabilities with the
scenarios obtained by the use of conventional simplified models, a actual data of the red zone, where all the fatalities were recorded,
comparison between the outcomes of the consequence analysis points out that the Viareggio accident should be considered among
discussed in Section 2.3 with the reported accident damages was the more severe scenarios taken into account in the approach.
carried out. The results are shown in Fig. 4.
Fig. 4a shows a comparison among the reported damage and the
maximum damage distance calculated for an instantaneous flash 5. Prevention of Viareggio-type accidents and mitigation
fire and a flash fire following an 80 mm continuous release and an measures
instantaneous release of the entire tank content (see Section 2.3
and Supplementary Material for input parameters definition). As A final issue concerning the analysis of the Viareggio accident in
shown in the figure, the extension of the red zone falls well inside the framework of TRA is the assessment of the effect of possible
the calculated maximum damage distances from the railway line. In prevention measures and mitigation barriers on the risk figures. A
particular, the damage distance corresponding to an instantaneous bow-tie methodology was used to assess the efficiency of several
release is far more extended than that experienced in the Viareggio protections to reduce the probability of the consequences of an
accident. Thus, the consequence-based maximum impact zones accident during LPG rail transport. The bow tie diagram presents
shown in Fig. 4a seem sufficiently conservative also when the relations between the causes of the accidents (left wing of the
compared to the Viareggio accident. It should be recalled that the bow tie), the resulting loss of containment (knot of the bow tie),
actual release section (about 150 mm equivalent diameter) caused and the relations with the possible consequences (right wing of the
a release rate that is reasonably between that assumed for a cata- bow-tie) (Delvosalle et al., 2006). It is worth mentioning that
strophic release and for an 80 mm release. among possible causes considered for the bow-tie analysis, inten-
tional acts of interference, such as external attacks to tankers (with
Fig. 4. Damage map and buffer lines related to different physical effects: a) solid line e ½ LFL concentration level following a continuous release from a 80 mm hole; dashed line e ½
LFL concentration level following an instantaneous release; b) solid line e 10% lethality, dashed line e 1% lethality, dashed-dotted line e 0.1% lethality.
G. Landucci et al. / Journal of Loss Prevention in the Process Industries 49 (2017) 36e46 43
explosives, shooting, or even impact with attackers' vehicles), effective maintenance. Finally, passive fire protections (complete
misoperations, or sabotages were not taken into account. Hence, installation of heat resistant coatings on tankers) or inherent safety
we focused the analysis on accidental transportation causes related measures, such as avoiding transport in populated areas or
to safety, excluding security issues. removing sharp objects near railway track, result in the highest
As mentioned in Section 2, the conventional release categories robustness level.
considered for LPG rail transportation accidents are basically two: From Table 8 it can be concluded that derailment detection and
instantaneous failure or continuous release. Instantaneous releases an emergency stop of the train is a good measure to reduce the
may in turn be divided in “hot” BLEVEs (caused by external fire) and probability of flash-fire, flares and BLEVEs, although not giving full
“cold” BLEVEs (not caused by fire) (Paltrinieri et al., 2009). An in- protection. Similarly, the removal of stakes from the vicinity of the
tegrated bow tie diagram is presented in Fig. 5. railway tracks (as the one that reasonably caused the breach in the
The numbered arrows present the relations between causes and tank) would be effective, due to high robustness, although not
final consequences. Safety measures can act as a line of defence and giving full protection. It is worth to notice that the stakes removal
interrupt the accident chain. The chain of events that took place in measure was already implemented 20 years ago in the Netherlands.
the Viareggio accident is presented in the central row of the bow- It is worth mentioning that several management strategies may
tie diagram. Table 8 lists the safety barriers that can be consid- significantly reduce the likelihood of accidents during hazmat
ered as defence lines, showing where the safety barrier acts as a line transport. As discussed by Liu et al. (2013), the combined adoption
of defence to reduce the likelihood of the final event. of tank car safety design improvement, deploying wayside defect
Most barriers cannot completely interrupt the accident chain. In detection technologies, upgrading track infrastructure, routing, and
some cases, the barriers act as a line of defence only in combination improving emergency response practices may contribute with
with other safety measures (e.g., a pressure relief valve can be interacting effects to the overall shipment risk reduction.
effective in combination with a heat resistant coating to prevent a A critical risk management strategy is related to the control and
hot BLEVE). limitation of train speed in order to reduce accident probability
Since there is not enough casuistry, experimental and literature during hazardous material transportation (Kawprasert and Barkan,
data to support a quantitative probabilistic assessment of the bar- 2010; Liu et al., 2013). It is worth to recall that in April 2015, after
riers, a qualitative reliability and robustness classification was the severe accident occurred in Lac-Me gantic, Canada (U.S.
performed based on to the literature on layer of protection analysis Department of Transportation, 2013), the Canadian Ministry of
(CCPS - Center of Chemical Process Safety, 2001). Transportation, perceiving “an immediate threat to safe railway
Procedural measures, e.g. barriers involving manual detection or operations”, issued an emergency directive (Canada Ministry of
improved inspection and maintenance, feature lower reliability, Transportation, 2015). The directive limits the speed of hazardous
since they strongly rely on human action. Active barriers, such as materials shipment in selected vulnerable zones from 80 to 65 km/
overfill protection, hot box or derailment detection, feature an in- h.
termediate level of reliability and robustness, since they require Concerning post release mitigation measures, a widely applied
external automatic activation, which might be subjected to strategy to limit the consequences of gas/vapor dispersion on
dormant failures. Hence these systems require periodic testing and population is the installation of mitigation walls or barriers (Busini
Fig. 5. Integrated bow tie diagram for an accident involving an LPG tank car. The central bold line evidences the development of the Viareggio accident.
44 G. Landucci et al. / Journal of Loss Prevention in the Process Industries 49 (2017) 36e46
Table 8
Safety barriers that may act as a line of defence for Viareggio-type accidents. Number under “line of defence” refers to barrier position in Fig. 5.
Measure Line of defence Flash fire Jet fire Hot BLEVE Cold BLEVE
Fig. 6. The mitigation wall installed on the East-side of Viareggio railway, evidencing the positioning respect to the damaged zone.
et al., 2012; Cormier and Mannan, 2008; Gilham et al., 2000; 6. Conclusions
Ponchaut et al., 2011; Tchouvelev et al., 2007). It is worth to
remind that the efficiency assessment of this type of protective The Viareggio LPG accident was analyzed in the perspective of
measures requires the detailed modeling of the obstacle influence HazMat transportation risk analysis. The results pointed out that
on the cloud dispersion (Derudi et al., 2014) and, in some condi- the Viareggio scenario, although of particular severity, is comprised
tions, instead of a mitigation effect the barrier might lead to a within those accounted in quantitative risk analysis. Thus, current
worsening effect of the dispersion scenario (Busini and Rota, 2014). tools used to assess risk figures and damage distances are still
A separation wall was built by the Viareggio Municipality few adequate even in the light of this severe accident, although the
months after the accident on the East-side of the railway lines in specific consequences require more detailed modeling tools to be
correspondence of the release point, as shown in Fig. 6. The wall is reproduced. Several safety barriers may be considered to control
2.5 m high and extends along the railway lines (South-North di- the risk due to these severe scenarios, even if the design of some
rection) for about 200 m, in a zone which was not provided with particular measures, such as mitigation walls, results critical for the
barriers before the accident (Fig. 3 evidences that only a hedge was effectiveness of the protection. The use of advanced tools as CFD
present). models is recommended when the detailed design of protection
A dedicated CFD study (Busini et al., 2011; Pontiggia et al., 2010) barriers is undertaken. The results also point out that QRA methods,
demonstrated that this measure is of poor effectiveness and the even if having a number of well-known limitations, may still be
consequences of the dispersion, given the same release scenario, useful to support decision making providing quantitative risk fig-
would have been of almost the same severity even with the current ures and a sound assessment of risk.
protection installed. This evidences the need of an accurate design More in general, the Viareggio accident, due to the derailment of
of passive protection systems and the introduction of advanced a train carrying hazardous materials among two Seveso sites, evi-
tools for the evaluation of safety issues in the framework of HazMat dences the need to consider the transportation of hazardous sub-
transportation. stances among the control of major accident hazards due to
The above discussion points out that the quantitative results dangerous substances addressed by Seveso Directives.
provided by QRA, being clearly related to physical elements of the
system, provide important information for the design of protection
systems, not available when qualitative methods based on all- Appendix A. Supplementary data
hazard indexes are applied.
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.jlp.2016.08.009.
G. Landucci et al. / Journal of Loss Prevention in the Process Industries 49 (2017) 36e46 45
References De Smedt, T., De Cremer, K., Vleminckx, C., Fierens, S., Mertens, B., Van Overmeire, I.,
Bader, M., De Paepe, P., Go €en, T., Nemery, B., Schettgen, T., Stove, C., Van
Oyen, H., Van Loco, J., Van Nieuwenhuyse, A., 2014. Acrylonitrile exposure in the
ACDS, 1991. Major Hazard Aspect of the Transport of Dangerous Substances. Advi-
general population following a major train accident in Belgium: a human bio-
sory Committee on Dangerous Substances. Health and Safety Commission,
monitoring study. Toxicol. Lett. 231, 344e351.
London, UK.
Delvosalle, C., Fievez, C., Pipart, A., Debray, B., 2006. ARAMIS project: a compre-
AEA Technology, 2001. Major Hazard Incident Data Service (MHIDAS).
hensive methodology for the identification of reference accident scenarios in
Alp, E., 1995. Risk-based transportation planning practice: overall methodology and
process industries. J. Hazard. Mater. 130, 200e219.
a case example. INFOR 33, 4.
Derudi, M., Bovolenta, D., Busini, V., Rota, R., 2014. Heavy gas dispersion in presence
Assadipour, G., Ke, G.Y., Verma, M., 2015. Planning and managing intermodal
of large obstacles: selection of modeling tools. Ind. Eng. Chem. Res. 53,
transportation of hazardous materials with capacity selection and congestion.
9303e9310.
Transp. Res. Part E Logist. Transp. Rev. 76, 45e57.
Direccion General de Transporte Ferroviario y Multimodal, 2007. Anuario estadís-
Batta, R., Chiu, S.S., 1988. Optimal obnoxious paths on a network: transportation of
tico ferroviario 2007, SCT e Secretarìa de Comunicaciones y Transportes
hazardous materials. Oper. Res. 36, 84e92.
[WWW Document]. http://www.sct.gob.mx/transporte-y-medicina-preventiva/
Birk, A.M., Cunningham, M.H., 1996. Liquid temperature stratification and its effect
transporte-ferroviario-y-multimodal/estadisticas-de-la-dgtfm/ (accessed
on BLEVEs and their hazards. J. Hazard. Mater. 48, 219e237.
01.07.10.).
Bonvicini, S., Antonioni, G., Spadoni, G., 2002. Road and Rail Transport of Dangerous
Egidi, D., Foraboschi, F.P., Spadoni, G., Amendola, A., 1995. The ARIPAR project:
Goods: Procedures, Tools and Guidelines for Risk Analysis and Routing, Report
analysis of the major accident risks connected with industrial and trans-
on Research Project No. 00.00610.PF37 Pos. 115.34713 Founded by the National
portation activities in the Ravenna area. Reliab. Eng. Syst. Saf. 49, 75e89.
Department of Civil Protection. University of Bologna, Bologna, Italy.
Erkut, E., Ingolfsson, A., 2005. Transport risk models for hazardous materials:
Boot, H., 2013. The use of risk criteria in comparing transportation alternatives.
revisited. Oper. Res. Lett. 33, 81e89.
Chem. Eng. Trans. 31, 199e204. http://dx.doi.org/10.3303/CET1331034.
Erkut, E., Verter, V., 1998. Modeling of transport risk for hazardous materials. Oper.
Brambilla, S., Manca, D., 2010. The Viareggio LPG railway accident: event recon-
Res. 46, 625e642.
struction and modeling. J. Hazard. Mater. 182, 346e357.
Erkut, E., Tjandra, S.A., Verter, V., 2007. Chapter 9 hazardous materials trans-
Brown, D.F., Dunn, W.E., 2007. Application of a quantitative risk assessment method
portation. In: Barnhart, C., Laporte, G. (Eds.), Transportation, Handbooks in
to emergency response planning. Comput. Oper. Res. 34, 1243e1265. http://
Operations Research and Management Science. Elsevier, pp. 539e621.
dx.doi.org/10.1016/j.cor.2005.06.001.
European Commission, 1982. Council Directive 82/501/EEC of 24 June 1982 on the
Bubbico, R., Di Cave, S., Mazzarotta, B., 2004a. Risk analysis for road and rail
major-accident hazards of certain industrial activities. Off. J. Eur. Communities
transport of hazardous materials: a simplified approach. J. Loss Prev. Process
L230, 1e18.
Ind. 17, 477e482.
European Commission, 1997. Council Directive 96/82/EC of 9 December 1996 on the
Bubbico, R., Di Cave, S., Mazzarotta, B., 2004b. Risk analysis for road and rail
control of major-accident hazards involving dangerous substances. Off. J. Eur.
transport of hazardous materials: a GIS approach. Loss Prev. Process Ind. 17,
Communities L10, 13.
483e488.
European Commission, 2003. Directive 2003/105/EC of the European Parliament
Bubbico, R., Maschio, G., Mazzarotta, B., Milazzo, M.F., Parisi, E., 2006. Risk man-
and of the Council of 16 December 2003 amending Council Directive 96/82/EC
agement of road and rail transport of hazardous materials in Sicily. J. Loss Prev.
on the control of major-accident hazards involving dangerous substances. Off. J.
Process Ind. 19, 32e38.
Eur. Communities L345, 97e105.
Busini, V., Rota, R., 2014. Influence of the shape of mitigation barriers on heavy gas
European Commission, 2006. Directive 2006/90/EC of 3 November 2006 Adapting
dispersion. J. Loss Prev. Process Ind. 29, 13e21.
for the Seventh Time to Technical Progress Council Directive 96/49/EC on the
Busini, V., Pontiggia, M., Derudi, M., Landucci, G., Cozzani, V., Rota, R., 2011. Safety of
Approximation of the Laws of the Member States with Regard to the Transport
LPG rail transportation. Chem. Eng. Trans. 24, 1321e1326.
of Dangerous Goods by Rail.
Busini, V., Lino, M., Rota, R., 2012. Influence of large obstacles and mitigation bar-
European Commission, 2012. . European Parliament and Council Directive 2012/18/
riers on heavy gas cloud dispersion: a liquefied natural gas case-study. Ind. Eng.
EU of 4 July 2012 on control of major-accident hazards involving dangerous
Chem. Res. 51, 7643e7650.
substances, amending and subsequently repealing council directive 96/82/EC.
Canada, S., 2011. CANSIM - Table Summary 404-0021-Rail Transportation, Origin
Off. J. Eur. Communities L197, 1e37.
and Destination of Commodities, Annual (Tonnes) [WWW Document]. http://
Eurostat, 2012. Annual Railway Transport of Dangerous Goods [WWW Document].
www5.statcan.gc.ca/cansim/a01?lang¼eng (accessed 24.02.11.).
http://epp.eurostat.ec.europa.eu/portal/page/portal/statistics/themes (accessed
Canada Ministry of Transportation, 2015. Emergency Directive Pursuant to Section
25.01.12.).
33 of the Railway Safety Act; Rail Transportation of Dangerous Goods [WWW
Fabiano, B., Curro , F., Reverberi, A.P., Pastorino, R., 2005. Dangerous good trans-
Document]. http://www.tc.gc.ca/eng/railsafety/railsafety-980.html (accessed
portation by road: from risk analysis to emergency planning. J. Loss Prev. Pro-
14.07.16.).
cess Ind. 18, 403e413.
Carter, D.A., Hirst, I.L., 2000. ‘Worst case’ methodology for the initial assessment of
Gheorghe, A.V., Birchmeier, J., Vamanu, D., Papazoglou, I., Kro €ger, W., 2005.
societal risk from proposed major accident installations. J. Hazard. Mater. 71,
Comprehensive risk assessment for rail transportation of dangerous goods: a
117e128.
validated platform for decision support. Reliab. Eng. Syst. Saf. 88, 247e272.
CCPS - Center for Chemical Process Safety, 2008. Guidelines for Chemical Trans-
Gilham, S., Deaves, D.M., Woodburn, P., 2000. Mitigation of dense gas releases
portation Safety, Security, and Risk Management. American Institute of
within buildings: validation of CFD modelling. J. Hazard. Mater. 71, 193e218.
Chemical Engineers - Center of Chemical Process Safety, New York, NY.
Guo, X., Verma, M., 2010. Choosing vehicle capacity to minimize risk for trans-
CCPS - Center of Chemical Process Safety, 1992. Guidelines for Hazard Evaluation
porting flammable materials. J. Loss Prev. Process Ind. 23, 220e225.
Procedures, second ed. American Institute of Chemical Engineers - Center of
Houdijk, R.M., 2016. Rail transport of hazardous substances from the perspective of
Chemical Process Safety, New York, NY.
“all hazard” risk management. Chem. Eng. Trans. 48, 949e954. http://
CCPS - Center of Chemical Process Safety, 1995. Guidelines for Chemical Trans-
dx.doi.org/10.3303/CET1648159.
portation Risk Analysis. American Institute of Chemical Engineers - Center of
HSE-Health Safety Executive, 1978. Canvey: an Investigation of Potential Hazards
Chemical Process Safety, New York, NY.
from Operations in the Canvey Island/Thurrock Area. HM Stationery Office,
CCPS - Center of Chemical Process Safety, 2000. Guidelines for Chemical Process
London, UK.
Quantitative Risk Analysis. American Institute of Chemical Engineers - Center of
HSE-Health Safety Executive, 1981. Canvey: a Second Report. A Review of the Po-
Chemical Process Safety, New York.
tential Hazards from Operations in the Canvey Island/Thurrock Area Three
CCPS - Center of Chemical Process Safety, 2001. Layer of Protection Analysis:
Years after Publication of the Canvey Report. HM Stationery Office, London, UK.
Simplified Process Risk Assessment. American Institute of Chemical Engineers -
Kawprasert, A., Barkan, C.P.L., 2010. Effect of train speed on risk analysis of trans-
Center of Chemical Process Safety, New York, NY.
porting hazardous materials by rail. Transp. Res. Rec. J. Transp. Res. Board 2159,
CFR Code of Federal Regulations, 2015. 49 CFR Part 179 “Specifications for Tank-
59e68.
Wagons.” U.S. Government Publishing Office, Washington, DC.
Landucci, G., Tugnoli, A., Busini, V., Derudi, M., Rota, R., Cozzani, V., 2011. The
CGSB - Canadian General Standards Board, 1997. CAN/CGSB-43.147e97. CGSB -
Viareggio LPG accident: lessons learnt. J. Loss Prev. Process Ind. 24, 466e476.
Canadian General Standards Board, Ottawa, Ontario, Canada.
Lees, F.P., 1996. Loss Prevention in the Process Industries, second ed. Butterworth -
Cormier, B., Mannan, M.S., 2008. LNG computational fluid dynamics consequence
Heinemann, Oxford.
modeling comparison with the Brayton Fire School tests. In: ACS National
Leonelli, P., Bonvicini, S., Spadoni, G., 1999. New detailed numerical procedures for
Meeting Book of Abstracts.
calculating risk measures in hazardous materials transportation. J. Loss Prev.
Cozzani, V., Antonioni, G., Bonvicini, S., Spadoni, G., Genovesi, P., Giovine, V., 2004a.
Process Ind. 12, 507e515.
La metodologia per la valutazione del rischio nel trasporto ferroviario di merci
Liu, X., Saat, M.R., Barkan, C.P.L., 2013. Safety effectiveness of integrated risk
pericolose. Argom. Ferrov. Ital. (Ital. Natl. Railw. Board) 120e137.
reduction strategies for rail transport of hazardous materials. Transp. Res. Rec.
Cozzani, V., Antonioni, G., Bonvicini, S., Spadoni, G., Genovesi, P., Giovine, V., 2004b.
102e110.
La metodologia per la valutazione del rischio nel trasporto ferroviario di merci
Maillette, J., Birk, A.M., 1996. Influence of release conditions on bleve fireballs. In:
pericolose (in Italian). In: VGR 2004 (Pisa, 19e21 October 2004. DIMNP Uni-
American Society of Mechanical Engineers. Pressure Vessels and Piping Division
versity of Pisa, Pisa, Italy, pp. 1e10.
(Publication) PVP, pp. 147e152.
Cozzani, V., Bonvicini, S., Spadoni, G., Zanelli, S., 2007. Hazmat transport: a meth-
Manca, D., Brambilla, S., 2010. Complexity and uncertainty in the assessment of the
odological framework for the risk analysis of marshalling yards. J. Hazard.
Viareggio LPG railway accident. J. Loss Prev. Process Ind. 23, 668e679.
Mater. 147, 412e423.
Manca, D., Brambilla, S., Totaro, R., 2010. A quantitative assessment of the Viareggio
46 G. Landucci et al. / Journal of Loss Prevention in the Process Industries 49 (2017) 36e46
railway accident. Comput. Aided Chem. Eng. 28, 187e192. perspectives. J. Transp. Eng. 119 (2), 177e188.
Milazzo, M.F., Lisi, R., Maschio, G., Antonioni, G., Bonvicini, S., Spadoni, G., 2002. Schmidt, J., 2012. Process and Plant Safety: Applying Computational Fluid Dy-
HazMat transport through Messina town: from risk analysis suggestions for namics. WILEY-VCH Verlag, Berlin, Germany.
improving territorial safety. J. Loss Prev. Process Ind. 15, 347e356. Schweitzer, L., 2006. Environmental justice and hazmat transport: a spatial analysis
Milazzo, M.F., Lisi, R., Maschio, G., Antonioni, G., Spadoni, G., 2010. A study of land in southern California. Transp. Res. Part D Transp. Environ. 11, 408e421.
transport of dangerous substances in Eastern Sicily. J. Loss Prev. Process Ind. 23, Spadoni, G., Leonelli, P., Verlicchi, P., Fiore, R., 1995. A numerical procedure for
393e403. assessing risks from road transport of dangerous substances. J. Loss Prev. Pro-
Ministere de l'Ecologie du De veloppement durable et de l'Energie, 2016. ARIA cess Ind. 8, 245e252.
(Analysis, Research and Information on Accidents) Database [WWW Docu- Spadoni, G., Egidi, D., Contini, S., 2000. Through ARIPAR-GIS the quantified area risk
ment]. http://www.aria.developpement-durable.gouv.fr/?lang¼en (accessed analysis supports land-use planning activities. J. Hazard. Mater. 71, 423e437.
11.02.16.). Spadoni, G., Contini, S., Uguccioni, G., 2003. The new version of ARIPAR and the
Ministero dei lavori Pubblici, 2001. Decreto Ministeriale 9 Maggio 2001, Suppl. Ord. benefits given in assessing and managing major risks in industrialised areas.
G. U. n.138 del 10/Giu/01, Requisiti minimi di sicurezza in materia di pianifi- Process Saf. Environ. Prot. Trans. Inst. Chem. Eng. Part B 81, 19e30.
cazione urbanistica e territoriale per zone interessate da stabilimenti a rischio Tchouvelev, A.V., Cheng, Z., Agranat, V.M., Zhubrin, S.V., 2007. Effectiveness of small
di incidente rilevante. barriers as means to reduce clearance distances. Int. J. Hydrogen Energy 32,
OPSI, 2009. The Carriage of Dangerous Goods and Use of Transportable Pressure 1409e1415.
Equipment Regulations and Come into Force on 1st July 2009. Tugnoli, A., Salzano, E., Di Benedetto, A., Russo, P., Cozzani, V., 2013. Analysis of the
Paltrinieri, N., Landucci, G., Molag, M., Bonvicini, S., Spadoni, G., Cozzani, V., 2009. flash-fire scenario in the viareggio accident. Chem. Eng. Trans. 32, 403e408.
Risk reduction in road and rail LPG transportation by passive fire protection. Uijt de Haag, P.A.M., Ale, B.J.M., 1999. Guidelines for Quantitative Risk Assessment
J. Hazard. Mater. 167, 332e344. (Purple Book). Committee for the Prevention of Disasters, the Hague (NL).
Ponchaut, N.F., Kyto €maa, H.K., Morrison, D.R., Chernovsky, M.K., 2011. Modeling the Unified Industrial & Harbour Fire Department Rotterdam-Rozenburg, 2016. FACTS,
vapor source term associated with the spill of LNG into a sump or impound- Failure and ACcidents Technical Information System. Unified Industrial &
ment area. J. Loss Prev. Process Ind. 24, 870e878. Harbour Fire Department. Rotterdam-Rozenburg, The Netherlands.
Pontiggia, M., Busini, V., Derudi, M., Alba, M., Scaioni, M., Rota, R., Landucci, G., U.S. Department of Transportation, 2012. Emergency Response Guidebook. US
Molag, M., Tugnoli, A., Cozzani, V., 2010. Safety of LPG rail transportation in the Department of Transportation, Pipeline and Hazardous Materials Safety
perspective of the Viareggio accident. In: Reliability, Risk and Safety: Back to the Administration, Washington DC.
Future, pp. 1872e1880. U.S. Department of Transportation, 2013. Lac-Me gantic Railroad Accident Discus-
Pontiggia, M., Landucci, G., Busini, V., Derudi, M., Alba, M., Scaioni, M., Bonvicini, S., sion and DOT Safety Recommendations. Report 4910-06-P. Washington, DC.
Cozzani, V., Rota, R., 2011. CFD model simulation of LPG dispersion in urban U.S. Department of Transportation, U.S. Department of Commerce, 2007. Hazardous
areas. Atmos. Environ. 45, 3913e3923. Materials - 2007 Economic Census e Transportation - 2007 Commodity Flow
Purdy, G., 1993. Risk analysis of the transportation of dangerous goods by road and Survey. EC07TCF-US(HM) [WWW Document].
rail. J. Hazard. Mater. 33, 229e259. Van Den Bosh, C.J.H., 1989. Green Book, Methods for the Determination of Possible
Reniers, G.L.L., Dullaert, W., 2013. A method to assess multi-modal Hazmat trans- Damage, CPR 16E. Committee for the Prevention of Disasters, the Hague (NL).
port security vulnerabilities: hazmat transport SVA. Transp. Policy 28, 103e113. Van Den Bosh, C.J.H., Weterings, R.A.P.M., 2005. Methods for the Calculation of
Reniers, G.L.L., Zamparini, L., 2012. History and Importance of Hazmat Trans- Physical Effects (Yellow Book), third. ed. Committee for the Prevention of Di-
portation, Security Aspects of Uni- and Multimodal Hazmat Transportation sasters, the Hague (NL).
Systems. Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany. Verma, M., 2009. A cost and expected consequence approach to planning and
Rhyne, W.R., 1994. Hazardous Materials Transportation Risk Analysis: Quantitative managing railroad transportation of hazardous materials. Transp. Res. Part D
Approaches for Truck and Train. Van Nostrand Reinhold, New York, NY. Transp. Environ. 14, 300e308.
RP Authority, 1982. COVO Risk Analysis of Six Potentially Hazardous Industrial Verma, M., 2011. Railroad transportation of dangerous goods: a conditional expo-
Objects in the Rijnmond Area, a Pilot Study. A Report to the Rijnmond Public sure approach to minimize transport risk. Transp. Res. Part C Emerg. Technol.
Authority. Central Environmental Control Agency, Schiedam, NL. 19, 790e802.
Ruffin, E., Cassir, C., Holvad, T., 2009. Impact Assessment on the Use of Derailment Verma, M., Verter, V., 2007. Railroad transportation of dangerous goods: population
Detection Devices in the EU Railway System, European Railway Agency Report exposure to airborne toxins. Comput. Oper. Res. 34, 1287e1303.
RA/REP/03e2009/SAF. European Railway Agency, Lille, France. VROM, 1996. Risk-criteria for the transport of hazardous substances. Ministry of
Saccomanno, F., Chan, A., 1985. Economic evaluation of routing strategies for haz- Housing, Planning and Environment, Ministry of Transport and Public Works,
ardous road shipments. Transp. Res. Rec. 1020, 12e18. the Hague (NL).
Saccomanno, F., Shortreed, J., 1993. Hazmat transport risks: societal and individual