A Risk Assessment Application of a Real Time Decision

Support System Model for HAZMAT Transportation in

a Sustainable Oriented Motorway Environment

Giampaolo Centrone

, Maria Pia Fanti

Gabriella Stecco

and Walter Ukovich

Dipartimento di Elettrotecnica, Elettronica ed Informatica

Universit`a degli Studi di Trieste, Italy

Dipartimento di Elettrotecnica ed Elettronica

Politecnico di Bari, Italy

Abstract. The transportation of hazardous material on congested motorways is

an area of increasing concern for public safety and environmental awareness. This

paper aims at developing a methodology with an original approach in making an

attempt to encompass both professional experience and theoretical knowledge

with application oriented studies from disparate areas related to the commercial

transportation of HAZMAT on motorway, intimately linked with the “sustainable

transportation” paradigm. The main objective is to assess quantitatively the ac-

ceptability of the Individual and Societal Risks connected with the transportation

of HAZMATs. In addition, we propose a real time model of a Decision Support

System for HAZMAT transportation on a sustainable oriented motorway envi-

ronment. Finally, we offer an application of the proposed model. The case study

involves a stretch of A4 motorway in the North-East of Italy.

1 Introduction

Economic globalization favors the increase of geographic mobility involving the ex-

pansion of transportation systems that is joined with the rise in land prices and the

increase of air and noise pollution. In this world development [13], dangerous goods

are used in many processes in industry all over the world and this has been justiﬁed

by the economic revenue generated by their use. A dangerous good (named hazardous

material or HAZMAT almost exclusively in the United States) is any solid, liquid, or

gas that can harm people, other living organisms, property, or the environment. Due

to its nature, every production, storage and transportation activity related to the use of

HAZMAT have many risks for both society and environment and are often subject to

chemical regulations. In this scenario, a new factor has acquired more and more impor-

tance: sustainability. Sustainability [5] is a systemic concept that relates to the continu-

ity of economic, social, institutional and environmental aspects of human society. As

HAZMATs are transported throughout the world in a great number of road shipments,

their commercial transportation could be catastrophic and poses risks to life, health,

Centrone G., Pia Fanti M., Stecco G. and Ukovich W. (2009).

A Risk Assessment Application of a Real Time Decision Support System Model for HAZMAT Transportation in a Sustainable Or iented Motorway

Environment.

In Proceedings of the 3rd International Workshop on Intelligent Vehicle Controls & Intelligent Transportation Systems, pages 37-46

 SciTePress

property, and the environment due to the possibility of an unintentional release. Trans-

portation of HAZMATs on road actually represents a potentially high risk in regard to

the nature of the HAZMAT carried by trucks and the physiochemical events associated

with these materials (radioactivity, explosion, toxicity, corrosion etc.), the localization

and the density of the concerned (population, economic activities, buildings, networks,

infrastructures, natural areas etc.), the characteristics and state of the roads (topogra-

phy, layout, tunnels etc.), the density of the trafﬁc, and the environmental conditions

(weather, natural events etc.). While HAZMAT accidents are rare events, in a sustain-

able vision of development it is necessary to integrate risk mitigation and prevention

measures into the transportation management in order to avoid the risks turning into

real events. In spite of this issue, HAZMAT type, quantity, itinerary and delivery time

are not precisely known by the public authorities, the highway and motorway compa-

nies, and the population. As a consequence, one of the main objectives of research in

this ﬁeld is to provide appropriate answers to the safety management of HAZMAT ship-

ments, in collaboration with the principal parties involved in the goods transportation

process. Researches in this area [6], focuses on two main issues: i) to assess the risk

induced on the population by HAZMAT vehicles traveling on the road network; ii) to

involve the selection of the safest routes to take.

1.1 Problem Deﬁnition

In Italy about 80% of road trafﬁc is represented by the delivery of goods, and the overall

trend in Europe seems to predict an increase of 30% within 2010. About 18% of this

freight trafﬁc is currently represented by HAZMAT transportation, but a clear aware-

ness of HAZMAT transportation world ﬂows on road and on the other transportation

modes - as well as of the related security and safety aspects - is not present yet, at least

from a social and economic point of view. Intelligent Transportation System technolo-

gies havealso made possible the gradual reduction in journey times and thus opening up

new economic horizons, with the conquest of wider markets. The freedom gained by the

ease of movement, however, has a cost in terms of environmental impact, quality of life

and safety. The risk is that the increasing demand for current and especially future can

make that the cost is no longer sustainable. However, the actual accident risk and impact

is not calculated. In addition, when, due to unforeseen events (trafﬁc jams, accidents,

etc.), they need to change route, they do not have any particular guidance on the safest

alternative route. Motorways are one of the most important supporting infrastructures

of transportation networks: they assure efﬁcient and safe mobility of persons and goods

in the world and represent the largest part of the built environment. Motorway is a term

for both a type of road and a classiﬁcation or designation. Motorways are high capacity

roads designed to carry fast motor trafﬁc safely. In the E.U. they are predominantly

dual-carriageway roads with a minimum of two lanes in each direction and all have

grade-separated access. Motorways are comparable with North American freeways as

road type, and interstates as classiﬁcation. In Italy, according to [3], HAZMATs trans-

portation by road should require constant monitoring (tracking and tracing) of vehicles

and cargo handled. This requirement involves a series of obligations to which must

meet companies under the European Agreement concerning the International Carriage

of Dangerous Goods by Road (ADR). As a consequence, motorway concessionaires

must adopt real time systems to monitor HAZMATs carried and support the decisions

on the transportation (MAS Monitoring - Alarm - Alerts).

2 Problem Solution

In Fig. 1 we propose a real time decision support system (DSS) model for monitoring

of HAZMAT vehicles, aiming at solving the above stated problems.

ALERTSISTEM

PMV

SMS

Radio

VVFF

PS

PC

INTERVENTION

PLANNING CRISIS

MANAGEMENT

RISK ASSESSMENT

INDIVIDUAL

SOCIAL

ALARP

thresholds

REAL-TIME MONITORING

Traffic situations

Weather conditions

Hazmat carried

SENSORS

Optical recognition

Meteorological-Station

Asim

GIS-distributed sensors

Historical

Incident Data

Hazmat

Data

Census and

population

Data

Operation

Procedures

Fig.1. The proposed System Architecture.

Such system should aim at calculating and evaluating in real time the individual

and societal risk related to the transit of HAZMAT on the motorway network. Then,

it should allow a monitoring in real time of the means transporting HAZMAT, a risk

assessment derived from the carriage, the alert and notiﬁcation of emergencies, and an

anomalies reporting for a subsequent planned intervention. The model derives from the

application of the quantitative risk assessment (QRA) methodology presented in Fig.

2(a).

We must take into consideration the following cause-effect chain which can be associ-

ated to a vehicle transporting one or more HAZMATs: the vehicle may be subject to

a road accident (accident); the accident may cause the release of material transported

(release); the release may cause a series of events (incident); the incident has an effect

in the area surrounding the point accident. The model refers to damage to persons and

in particular to death. The model of risk assessment derived from road transportation of

HAZMAT is presented by a schematic representation in Fig. 2(b).

Risk assessment is typically structured as a process resulting from the interaction among

the transportation network (in this case motorway), the vehicle (or better the traveling

risk source), and the impact area. The model evaluates simultaneously the consequences

and the frequencies of occurrence of possible scenarios. This makes it possible to assess

Transportation Risk Analysis

Network

characterization

Utilization of risk

estimation

Risk estimation

Consequence

estimation

Likelihood

estimation

Release/scenario

analysis

(a) The Transport Risk Assessment

Methodology.

Impact on area

accident

Potential social

damage

Social Risk

Individual

damage

Individual Risk

Dangerous

substance release

Accident

Hazmat

Incident

(explosion,

expanding

vapor, jet fire,

etc.)

Parameters:

transportation network

mobile source of risk

geographical area

(b) The Risk Assessment Model.

Fig.2.

quantitatively the individual risk and the societal risk (if the distribution of the people

liable to be exposed is at hand). A complete assessment of the risks due to HAZMAT

by road would require to consider all the possible weather situations, all the kinds of

accidents, with all the types of vehicle partially or fully loaded. Such an evaluation

is completely impossible and some simpliﬁcations have to be introduced. The QRA

model is based on the following steps: choice of a restricted number of HAZMATs,

choice of some representative accidental scenarios implying those HAZMAT with their

usual packagings, identiﬁcation of physical effects of those scenarios for an open air

or a tunnel section, evaluation of their physiological effects on road users and local

population, taking into account the possibilities of escape/sheltering, determination of

the yearly frequency of occurrence for each scenario.

F (N) curves and their expected values are the major outputs of the QRA model. Ac-

cording to [7], Frequencies / Gravity curves (F (N) curves) stand for the annual fre-

quency of occurrence F to have a scenario likely to cause an effect (generally, the

number of fatalities) greater than or equal to N and Expected value (EV) is the number

of fatalities per year, obtained by integration of a F (N) curve.

3 How to Characterize the Risk in Transporting of HAZMAT on

Motorway Environment: A Case Study

In this section we present an application of the real time model for the calculation of

the Individual and Societal Risks in a motorway environment.

The Transportation Network. We consider the motorway network of an Italian con-

cessionaire, in the North-East of Italy, S.p.A. Autovie Venete. In particular we refer to

the sections summarized in Tab. 1. The network under consideration has been modeled

as a network with nodes and links where nodes represent exits / junctions of the motor-

way and links the stretches (sections) of motorway between two exits / junctions. Let

link

be the set of links of the motorway network and l a generic link. For each link,

Table 1. Link length, average population density, HAZMAT and number of vehicles carrying

them on tested links.

Link Length Average Population Substance type Num. of

(Km) Density (inhab/Km

) Vehicles

S.Stino di Livenza - Portogruaro 12.8 214.35 Chlorine 2

Portogruaro - Latisana 13.5 166.05 Ammonia 2

Latisana - S. Giorgio di Nogaro 17.6 112.66 Hydrochloric Acid 2

Nitric Acid 2

the following data have been obtained: length [Km], and average population density [in-

habitants / Km], around the link. According to [18], the average population density has

been calculated using a GIS (Geographical Information System) overlapping the geo-

graphical map of the municipalities on the motorway network of S.p.A. Autovie Venete

in order to identify common cross on each link. Then, for each link, we have identi-

ﬁed the municipalities involved and measured the kilometers of infrastructure that pass

through each town in order to identify the weights for calculating the average density

on the link. These weights have been derived by dividing the kilometers of infrastruc-

ture that affect each municipality with the total length of the link. Note the density of

population in each Italian municipality, using data on the census of 2001 [9], we shall

calculate the weighted average with weights determined in the previous step. Tab. 1

illustrates the links in discussion with the relevant data.

The Accident Probability. We use the Truck Accident Rate of Harwood [8] in order

to calculate the accident probability (λ

inc

(l)) in terms of e vents/(vehicle ∗ km). We

can also calculate the rate of accidents on a single stretch of length unit road by using

the number of accidents in a time period of ten years and the total distance traveled

by heavy vehicles during the same period, data provided by AISCAT (Associazione

Italiana Societ´a Concessionarie Autostrade e Trafori) [1].

T AR

V KT

(1)

where T AR

is the average accident rate for trucks events/(vehic le∗ km) on the Ital-

ian motorway network for year y

; A

is the number of accidents involving trucks on

the Italian motorway network; V KT

is the total distance traveled(vehicle-kilometers)

by trucks on the network under consideration. Tab. 2 shows the number of accidents in-

volving heavy vehicles, the total distance traveled and the Truck accident rate year by

year from 1997 to 2007 and the summary data (extension, routes) for the years under

consideration. The last row presents the data that we use in the model.

The HAZMAT. in the next step, we select the HAZMATs that will be considered in

calculating the risk. In particular, according to [16], we have considered substances that

are more frequent or signiﬁcant on the network under consideration. For each of these

goods we obtained the following data (summarized in Tab. 3):

– the probability of release due to the accident (p

rel

(ν)) depending in general on the

characteristics of the vehicle transporting the HAZMAT and on the type of accident

where the vehicle is involved. This probability is taken from [2].

Table 2. Accidents and summary data year by year from 1997 to 2007.

ROUTES ACCIDENTS TAR

YEAR EXTENSION (veh. − km)

KM Heavy Heavy Heavy

1997 5371 1.4428∗10

7825 5.42∗10

−7

1998 5380 1.5161∗10

8854 5.84∗10

−7

1999 5380 1.5974∗10

10024 6.28∗10

−7

2000 5380 1.6790∗10

9681 5.77∗10

−7

2001 5388 1.7254∗10

9647 5.59∗10

−7

2002 5388 1.7836∗10

9691 5.43∗10

−7

2003 5388 1.8359∗10

9198 5.01∗10

−7

2004 5391 1.9059∗10

8841 4.64∗10

−7

2005 5432 1.9184∗10

9005 4.69∗10

−7

2006 5441 1.9764∗10

9000 4.55∗10

−7

2007 5446 2.0229∗10

8613 4.26∗10

−7

1.9403∗10

100379 5.17∗10

−7

– the types of possible releases classiﬁed in relation to the size of the leakage hole

rel,t

(ν)) and the rate of release or the amount of material spilled (p

rel,t

– the types of consequences of incident caused by different types of release of HAZ-

MAT given the accident for a given type of substance (N

out

(ν, r)).

– the likelihood of occurrence of a ﬁnal result given the incident (p

out

(i)). This prob-

ability is derived for each triplet [substance - leakage - type of ﬁnal outcome] from

the information relating to incidents involving HAZMAT from 1997 to 2008 re-

ported in the HMIS database [14]. This database contains detailed information on

accidents involving HAZMAT in the U.S..

– the frequency of occurrence of a given scenario: f

scen

ν,t

(i, r) = λ

inc

· p

rel

(ν) ·

rel,t

(r) · p

out

(i)

– the lethal area radius of each pair [type of release - ﬁnal outcome] calculated using

the free software RMPComp distributed by U.S. EPA (Environmental Protection

Agency) [15].

Assumptions. In the application we have considered the following assumptions.

1. λ

inc

(l) uniform throughout the link and constant for all links taken into considera-

tion: λ

inc

;

2. exposure area of danger circle type [7] centered at the point of the incident with a

radius depending on the type of substance, release and ﬁnal outcome;

3. any person within the exposure area suffers from the same injury (death) in the same

way regardless of the position, while people outside that area are not affected;

4. the seasons (j), the weather conditions on link C

met

(l) and the wind direction ϑ

(l)

are not taken into account;

5. the simulation is performed on a single moment in time;

6. risk neutral model (α = 1) [17].

Individual Risk Calculation. The simulation was carried out on three adjacent links.

As individual risk is the annual probability of an individual placed in a designated point

Table 3. Frequency scenarios.

HAZMAT p

rel

Release

Type

(Spillage)

relt

Incident

Type

(Cloud)

out

Incident

Prob.

Scenario

Frequency

Lethal

area

radius

(Km)

Chlorine 0.010 Small 0.94 Toxic 1 9.40∗10

−3

4.86∗10

−9

1.0

0.010 Medium 0.04 Toxic 1 4.00∗10

−4

2.07∗10

−10

2.8

0.010 Large 0.02 Toxic 1 2.00∗10

−4

1.03∗10

−10

5.6

Ammonia 0.025 Small 0.93 Toxic 1 2.31∗10

−2

1.20∗10

−8

0.2

0.025 Medium 0.05 Toxic 1 1.33∗10

−3

6.88∗10

−10

1.0

0.025 Large 0.02 Toxic 1 5.32∗10

−4

2.75∗10

−10

2.1

Nitric Acid 0.015 Small 0.93 Toxic 1 1.39∗10

−2

7.19∗10

−9

0.3

0.015 Medium 0.06 Toxic 1 8.82∗10

−4

4.56∗10

−10

0.5

0.015 Large 0.01 Toxic 1 2.21∗10

−4

1.14∗10

−10

1.9

Hydrochloric Acid 0.015 Small 0.92 Toxic 1 1.38∗10

−2

7.13∗10

−9

0.3

0.015 Medium 0.05 Toxic 1 7.37∗10

−4

3.81∗10

−10

0.8

0.015 Large 0.03 Toxic 1 4.81∗10

−4

2.49∗10

−10

2.6

Table 4. Geographical coordinates of the points chosen for Individual Risk calculation.

Point Location Latitude Longitude

Portogruaro Centro 45.78 12.83

Area di Servizio Fratta Nord 45.80 12.88

Latisana Ospedale 45.77 13.00

Muzzana del Turgnano Centro 45.82 13.13

of interest is affected by some degree of damage as a result of a speciﬁc incident [10],

four points were chosen as “hot spots” at which to calculate the individual risk. Tab.

1 and 4 show respectively the links with the relevant substances circulating and the

geographical coordinates of the points chosen for the calculation of individual risk. The

network portion and the points under consideration are represented in Fig. 3.

For the calculations we have used the formula presented in [11] and [12] suitably mod-

iﬁed to take into account the previous assumptions, the motorway environment and the

real time events. More details can be found in [4]. Consequently, we made explicit that

each event i belongs to the N

(out)

(ν) of the general model may consist of a pair [type

of release - ﬁnal outcome] as in this case.

IRP =

links

i=1

veh

(l)

v=1

rel

(ν)

r=1

type

(l, ν)f

rel

(ν, r)·

out

(ν,r)

i=1

out

(i) · V

Q(x)v→S

(i)dL

(2)

rel

(ν, r) = λ

inc

· p

rel

(ν) · p

rel,t

(r) (3)

Fig.3. Representation of points and the network portion under consideration.

Table 5. Simulation results - Individual Risk.

Point Location Individual Risk

Portogruaro Centro 2.28∗10

−9

Area di Servizio Fratta Nord 5.78∗10

−9

Latisana Ospedale 0

Muzzana del Turgnano Centro 1.75∗10

−9

where N

veh

(l) is the number of different vehicle topologies on link l, N

type

(l, ν) is

the number of vehicles carrying the dangerous substance ν currently in transit on the

link l, N

rel

(ν) is the number of release cases of the dangerous substance ν, L

is the

route of link l and V

Q(x)

→S

(i) is equal to 1 if the point S is inside the danger circle

centered at the point of possible accident Q related the triplet [substance - release type

- ﬁnal outcome]; 0 if the point S is external the same danger circle. Line integral was

calculated using the method of Cavalieri-Simpson, dividing each of the three links in

10 intervals of equal length. Tab. 5 shows the results of the simulation.

From the evidence we can establish that the individual risk in the four points is accept-

able according to the British ALARP threshold as the value is much lower than the limit

value of 1 0

−6

[10].

Societal Risk Calculation. In order to calculate the societal risk, we refer again to [11]

and [12] suitably modiﬁed. For each link knowing the vehicles that are going through,

we use (4) to obtain the F (N) curves representation (for details see [4]).

F (N) =

link

i=1

veh

(l)

v=1

type

(l, ν) ·

out

(ν)

i=1

scen

(i, C

met

(l))dL

(4)

Number of deaths

F(N) curve Upper limit

Lower limit

Denmark U.K. Holland

Fig.4. Simulation results - Societal Risk.

The curve F (N ) is drawn in Fig. 4, referred to the simulation with different thresholds

of acceptability [12].

It can be seen that at the moment of the simulation the societal risk, according to

the British acceptability thresholds, is in the ALARP zone, whereas according to Dutch

and Danish thresholds it is not acceptable.

4 Conclusions

The transportation of HAZMAT on congested motorways is becoming an area of in-

creasing concern for public safety and environmental awareness. The risk to population

and damage to environment is a major concern to the general public and government

policy makers. Against these problems we present a methodology to perform the indi-

vidual and societal risk assessment related to HAZMAT transportation in a sustainable

oriented motorway environment. It constitutes an approach based on the GIS. The as-

sessment criteria, based on the “sustainable transportation” paradigm, are structured

into efﬁciency, cohesion and environmental criteria. The aim is assessing whether these

risks are acceptable and possibly, if they were not, notify the situation through alert

messages in order to take appropriate actions. We offered an application of the pro-

posed real time model for the calculation of the Individual and Societal Risks involving

in the case study a stretch of A4 motorway in the North-East of Italy. In spite of a

limited number of trucks transporting HAZMAT on the motorway, the results of the

application point out the concrete possibility to exceed the thresholds of the ALARP

limits for the societal risk.

With regard to possible developments, the QRA methodology could be to extend, in

particular the model for calculating the individual and societal risk, to other situations

of HAZMAT transportation by other transportation modes.

References

1. AISCAT: Rapporto incidentalit´a rete autostradale italiana 1997-2007. Statistiche Aiscat, As-

sociazione Italiana Societ´a Concessionarie Autostrade e Trafori, (2008)

2. Brown, D.F., Dunn, W.E.: Application of a Quantitative Risk Assessment Method to Emer-

gency Response Planning. Computers & Operations Research, 34 (2007) 1243–1265

3. CEI: Road trafﬁc and transport telematics Standards framework. Review of the applications

and technological trends, normative references Annex to CEI UNI 70031 (2001) 1–18

4. Centrone, G.: Modeling a real time decision support system for HAZMAT transportation in

a sustainable oriented motorway environment. Ph.D. Thesis, Universit´a di Trieste (2009)

5. Centrone, G., Coslovich, L., Pesenti, R., Ukovich, W.: Social responsibility and sustainability

in motorway Corporate Governance. International Journal of Environment and Sustainable

Development 7 (2008) 94–113

6. Centrone, G., Pesenti, R., Ukovich, W.: Hazardous Materials Transportation: a Literature

Review and an Annotated Bibliography. In: Bersani, C., Boulmakoul, A., Garbolino, E.,

Sacile, R. (eds.): Advanced Technologies and Methodologies for Risk Management in the

Global Transport of Dangerous Good. NATO Science for Peace and Security Series, Vol. 45.

IOS Press, Amsterdam (2008) 33–62

7. Erkut, E., Tjandra, S.A., Verter, V.: Hazardous materials transportation. In: Barnhart, C.,

Laporte, G. (eds.): Transportation, Handbook in OR & MS, Vol 14. Elsevier (2007) 539–621

8. Harwood, D.W., Russell, E.R., Viner, J.G.: Procedure for developing truck accident and re-

lease rates for hazmat routing. Journal of Transportation Engineering 119 (1993) 189–199

9. ISTAT: Quattordicesimo censimento generale della popolazione italiana (2001).

http://cartema2.istat.it/home cartema/iisstart.htmh

10. Jonkman, S.N., van Gelder, P.H.A.J.M., Vrijling, J.K.: An overview of quantitative risk mea-

sures for loss of life and economic damage. Journal of Hazardous Materials 99 (2003) 1–30

11. Leonelli, P., G. Spadoni, G.: A new numerical procedure for calculating societal risk from

road transport of dangerous substances. In Proceedings SRAEurope Annual Meeting (1996)

243–246

12. Leonelli, P., Bonvicini, S., Spadoni, G.: Hazardous materials transportation: a Risk-analysis-

based routing methodology. Journal of Hazardous Materials 71 (2000) 283–300

13. Nederveen, A.A.J., Konings, J.W., Stoop, J.A.: Globalization, International Transport And

The Global Environment: Technological Innovation, Policy Making And The Reduction Of

Transportation Emissions. Transportation Planning and Technology 26 (2003) 41–62

14. PHMSA - U.S. Department of Transport: HMIS database. (2006)

https://hazmatonline.phmsa.dot.gov/IncidentReportsSearch/Search.aspx

15. RMP*Comp 1.07 http://www.epa.gov/OEM/content/rmp/rmp comp.htm

16. Rossetto, A.: Sviluppo di un sistema DSS per il monitoraggio e l’analisi quantitativa del

rischio nel trasporto di merci pericolose in autostrada. MSc. Thesis, Universit´a di Trieste

(2008)

17. Slovic, P.: Perception of risk. Science (1987) 280–285

18. Zhang, J.J., Hodgson, J., Erkut, E.: Using GIS to assess the risks of hazardous materials

transport in networks European Journal of Operational Research 121 (2000) 316–329