- Email: [email protected]
Preliminary study on the transport of hazardous materials through tunnels
Accident Analysis and Prevention 41 (2009) 1199–1205
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Preliminary study on the transport of hazardous materials through tunnels Roberto Bubbico, Sergio Di Cave, Barbara Mazzarotta ∗ , Barbara Silvetti Dipartimento di Ingegneria Chimica Materiali Ambiente, Universit` a di Roma “La Sapienza”, Via Eudossiana 18, 00184 Rome, Italy
a r t i c l e
i n f o
Article history: Received 25 September 2007 Received in revised form 4 April 2008 Accepted 25 May 2008 Keywords: Tunnel Transport Hazardous materials Risk analysis
a b s t r a c t The risk associated to road and rail transportation of some hazardous materials along two routes, one including a significant portion in tunnels, and the other following the same path, but running completely in the open, is assessed. The results show that, for rail transport, no particular risk increase or mitigation is associated to the circulation of the dangerous goods through tunnels; on the contrary, for road transport, a risk increase is generally observed in the presence of tunnels. However, for LPG, the risk curve in the open lies above that in tunnels in the high frequency–low fatality zone, according to the different evolution of the accidental scenarios in the tunnel (assuming no ventilation). The transportation of liquefied nitrogen, not hazardous in the open but potentially asphyxiating in a tunnel, gives rise to a negligible risk when performed by rail, but to a not negligible one, when performed by road. These preliminary results focused on the risk for the exposed population, suggest that it may be unnecessary to limit dangerous goods circulation through rail tunnels, while, at least for some types of dangerous goods, the circulation through road tunnels may be allowed/forbidden based on the results of a specific risk analysis. © 2008 Elsevier Ltd. All rights reserved.
1. Introduction Long distance road and rail connections in Europe often include the passage through tunnels belonging to the National networks or connecting different Countries. Both these situations occur in Italy, with the long mountain chain of Apennines running from the north–west to the south–east of the Country, and with the Alps as boundary separating Italy from France, Switzerland, Austria and Slovenia. As a matter of fact, in Italy 83 road tunnels and 255 rail tunnels exceed 1 km in length; presently, three rail and three road tunnels longer than 10 km exist through Alps, with two more planned, whose length will exceed 50 km. A number of severe tunnel accidents occurred in the last years in Europe, with a high toll in terms of human lives, as well as direct costs for repairing the infrastructure and indirect costs associated with the temporary closure of the route including the gallery. It has to be remarked that, indeed, hazardous materials were not involved in any of these accidents; nevertheless, even common freight and passenger vehicles demonstrated their potential for generating a hazardous accidental scenario inside a tunnel, where the confinement causes longer persistence and higher levels of temperature, of concentration of toxic gases, and of overpressure with respect to
∗ Corresponding author. Tel.: +39 06 4458590; fax: +39 06 4827453. E-mail addresses: [email protected] (R. Bubbico), [email protected] (S. Di Cave), [email protected] (B. Mazzarotta), [email protected] (B. Silvetti). 0001-4575/$ – see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.aap.2008.05.011
open air. Moreover, people inside a tunnel may experience difficulties when trying to escape due to poor visibility, distant location of safe places and difficulty to locate them, and stressing environmental conditions. The need for increasing tunnel safety requirements is well present both to Public Authorities and to the general public, and some specific provisions are being issued for long rail and road tunnels. 1.1. Rail tunnels A group of experts from the United Nations Inland Transport Committee (United Nations, 2003) has studied the safety of tunnels in the range 1–15 km (for tunnel exceeding 15 km specific additional safety measures are enforced). The outline of the guidelines is that of requiring a basic, simplified, risk analysis for tunnels ranging from 2 to 9 km, and a standard risk analysis for tunnels exceeding 9 km. A regulation issued by the Italian Ministry of Infrastructures and Transportation (MIT, 2005), concerning safety requirements for rail tunnels, considers all the aspects involving infrastructure, exercise, users, and carriages. The requirements are listed according to the following objectives: • Incident prevention; • Consequence mitigation;
1200
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
• Evacuation of the passengers; • Emergency relief. A set of minimum requirements applies to all rail tunnels, and no risk analysis is needed for tunnels where: • length does not exceed 2 km; • railway traffic is less than 220 trains/day; • trains carrying hazardous materials and passenger trains cannot simultaneously circulate in the tunnel; • there is no slope inversion in the tunnel; • there are no specific area risks close to the tunnel inlets. Risk analysis has to be performed for all tunnels not fulfilling the previous requirements, and additional provisions may be set to grant a sufficient safety level. The risk analysis procedure is specified, and the basic scenario is a fire in a car of a passenger train. Acceptability thresholds are set for both individual and societal risk: for example, the ALARP limits are 1 × 10−9 –1 × 10−11 events per passenger and per km for individual risk. 1.2. Road tunnels The European Directive 2004/54/EC (CD, 2004) sets the minimum safety requirements of road tunnels exceeding 500 m, according to a two-steps approach: 1. a minimum safety level will be granted to all road tunnel users; 2. the sustainability of the present transportation system, especially in mountainous districts, will be verified, possibly adopting alternative transportation modes and routes. Accident prevention and mitigation of the consequences are obtained by adopting proper management and technical standards. The organizing aspects define jurisdiction and tasks of some key figures: • Administrative authorities, with the general responsibility of tunnel safety at a national, regional or local level. • Inspection services, assessing tunnel safety by means of tests and inspections. • Tunnel manager, responsible for tunnel exercise. • Safety managers, responsible for implementing all tunnel prevention and safety measures. Technical requirements concern: • Infrastructures, classified in five levels, depending on tunnel type, volume, traffic and length. To each level specific features of the tunnel (number of pipes, escape routes, ventilation, emergency exits, and possibly, special provisions) should correspond. A minimum set of requirements concerning escape routes, extinguishers, radio broadcasting, video-surveillance, cabling specific road signals to indicate escape routes and safe place location will apply to practically all tunnels. • Rules concerning works in tunnels, incidents management, activities of the control center (when present), tunnel closure, hazardous materials transportation, vehicle passing and minimum vehicle distance. • Vehicles, with provisions concerning the presence of extinguishers on the vehicles, the prohibition of filling additional fuel tanks and, possibly, additional safety measures. • Information to the users, issuing regular informative campaigns about tunnel safety, instructing the public about proper behavior in case of incidents and the location of safety devices.
Most existing tunnels have been designed and constructed some decades ago, and it is well recognized that the costs for their adaptation to these more demanding standards may be extremely high: for this reason a 10 years deadline has been set, which may be extended 5 years further in the Countries with a particularly high presence of tunnels. 1.3. Dangerous goods circulation in tunnels Generally speaking, road and rail transportation of hazardous materials is ruled by ADR and RID codes, respectively (CD, 1994, 1996); however, these documents do not provide any indications related to the passage of vehicles carrying dangerous goods through tunnels. The guidelines cited in Section 1.1 (United Nations, 2003) suggest that, based on the results of risk analysis, it may be also possible to prohibit the simultaneous passage of hazardous goods and passenger trains in tunnel longer than 2 km. Similarly, the regulation issued by the Italian Ministry of Infrastructures and Transportation (MIT, 2005), as far as dangerous goods transportation is concerned, relies only on operating procedures for preventing incidents. In particular, trains carrying hazardous materials and passenger trains will not be allowed to simultaneously circulate in the tunnel, which will be obtained by adopting a proper scheduling of the trips or limiting the transit to dangerous goods. This lack of general rules persists also in the case of road tunnels. As a matter of fact, a study concerning road tunnels (OECD, 1997) showed that no general code exists, and also in the European Directive previously cited in Section 1.2 (CD, 2004), no specific reference is made to hazardous materials circulation in tunnels, even if the general measures suggested may be useful also in such a case. Presently, however, regulations concerning hazardous materials circulation are issued only for specific (long) tunnels and, especially, for those connecting different Countries: the circulation of hazardous materials in the tunnel may be free, limited to specific ADR/RID products, or forbidden. By examining the actual situation for some long European tunnels, it seems that no particular restrictions apply to rail circulation of RID materials (with the only exception of the Eurotunnel below the English Channel), while, for the road case, the circulation may be prohibited to some classes of ADR materials (tunnel of Frejus) or to all of them (tunnel of Mont Blanc). The issue concerning the circulation of dangerous products is well present also from the planning stage of two new European tunnels (56 km between Innsbruck and Brenner, and 54 km on the new rail line connecting Lyon and Turin). It is clear that the subject concerning the circulation of hazardous materials in tunnels is rather complex and that any sound approach to the problem should be based on risk analysis. In fact, just forbidding dangerous goods circulation, from one hand will increase the safety of tunnel users, but, on the other hand, will submit to additional risks the population living close to or travelling along the alternative routes. In fact, it should be born in mind that alternative paths avoiding tunnels are generally much longer than direct routes, may present higher accident rates, and, above all, may cross a number of urban sites where the consequences of an accident may affect a greater number of people with respect to tunnel users. In this paper, a transportation risk analysis (TRA) approach is applied to assess the risk associated to the transport of some different hazardous materials by road and by rail along routes including significant portions inside tunnels. The results are compared with those obtained assuming the same routes to run completely in open air, in order to preliminarily assess the extent of risk increase/decrease to be expected for each product and transport modality.
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
2. Procedure Transportation risk analysis (TRA) concerning hazardous materials is basically an extension of the well-known (CCPS, 2000) quantitative risk analysis (QRA). In particular, it takes into account that the risk source is not fixed, but travels in a continuously changing environment, by dividing the route into homogeneous portions where all the parameters involved in risk calculations (accident rate, scenario probability, population at risk, etc.) can be assumed to keep constant (CCPS, 1995). The risk is usually assessed in terms of societal risk, which requires to estimate: • the frequency of all the final outcomes deriving from the accidental scenarios; • the extension of the areas where the consequences of the final outcomes may cause fatalities to the exposed population; • the expected number of people in those impact areas. Accordingly, the information concerning the transport case (product amount and physical state, number of trips, seasonal/daily schedule of the trips) should be combined with those related to each route portion, and consequence analysis calculations have to be carried out to account for the different scenarios, outcome cases and environmental conditions. Specific software tools, often based on a GIS (Geographic Information System) to manage route-related information, have been developed to assist TRA (Bubbico et al., 2004). 2.1. Open air and tunnel sections A number of differences may be encountered when applying TRA to open air and tunnel portions of a route: • Accident rate: the different environmental conditions experienced inside a tunnel (for example, poor visibility, but also the absence of snow, fog and rain), as well as the possible enforcement of different speed limits or safety distances, may affect the accident rate. Accordingly, for road transport, the expected accident rate may increase, as well as decrease, with respect to an open air section; on the contrary, for rail transport, the accident rate is not expected to vary significantly in tunnels. • Accidental scenarios and final outcomes: the presence of a confinement inside a tunnel causes the accidental scenarios to deeply change. As a limit case, the release of liquefied nitrogen, which is an inert product, can be considered: in open air the spill will evaporate and disperse safely; in a tunnel, on the contrary, the sudden evaporation of a large amount of nitrogen may reduce local oxygen concentration, and possibly cause the death of the exposed population. • Given an accidental scenario, the range of possible final outcomes also changes: for example, the release of a flammable material in a tunnel, in case of ignition, besides thermal radiation (which is practically the only effect in the open) can generate a toxic, and hot, smoke cloud rapidly invading the tunnel. • Finally, the probability associated to each branch of the event tree, describing the evolution of the accidental scenario into the various final outcomes, may be different in the open and inside tunnels. For example, due to the confinement, a VCE is more likely to occur in a tunnel rather than in the open. • Consequences of the final outcomes: the extension of the impact area changes remarkably in a tunnel with respect to open air. Thermal radiation levels are higher, due to confinement, and more persistent, due to the thermal insulation provided by the tunnel structure. Peak overpressures increase, too, due to the (partial)
1201
reflection of the shock waves operated by tunnel walls and vault. Toxic substances concentrations reach higher values, and the toxic clouds may easily pervade the whole section of the tunnel, in case of both buoyant and dense clouds, and for an extension of many hundreds meters. • People in the impact area: in the open both residents and on-route population can be affected by the consequences of the release of a hazardous material, while, in a tunnel, only on-route population is at risk. However, in the open, people can easily escape from the accident location, which is not true in a tunnel. For the scope of this preliminary study, the differences possibly occurring in the accident rates between open air and tunnel sections of the route have been disregarded, and the attention has been focused on accidental scenarios, and on outcome cases probabilities and consequences. 2.1.1. Accidental scenarios and final outcomes The accidental scenarios for hazardous materials transportation usually consist in the spill of the product from holes of different sizes, ranging from a puncture to a catastrophic rupture capable of rapidly emptying the container. Generally, TRA in the open takes into account the release of flammable and toxic liquid or gases; in the case of tunnels, in addition to these products, inert gases (both gaseous and liquefied) have to be considered, since their release may result asphyxiating in a confined environment. The final outcomes assumed in the open depend on the physical state and on the characteristics of the hazardous material: • • • •
Flammable liquid: pool fire, flash fire, VCE. Flammable gas: jet fire, flash fire, VCE. Flammable liquefied gas: jet fire, flash fire, VCE, BLEVE, fireball. Toxic gas or liquid: toxic cloud.
For a tunnel, due to the containment, the final outcomes are assumed as follows (OECD, 2003): • Flammable liquid: pool fire, flash fire, VCE, toxic cloud. • Flammable gas: jet fire, flash fire, VCE, toxic cloud. • Flammable liquefied gas: jet fire, flash fire, VCE, BLEVE, fireball, toxic cloud. • Toxic or inert gas or liquid: toxic cloud. The models to be used for determining the heat radiation, overpressure and concentration profiles in the open and inside a tunnel are also different. In the former case a number of mathematical models are reported in the literature (CCPS, 2000) and consequence analysis software are available; in the latter case, the proposed approaches may be based either on very simplified models, like those proposed by the international group of experts working on tunnel safety (OECD, 2003) or on very demanding mathematical modelling, performed by means of Computational Fluid Dynamics (CFD). Another difference between open air and tunnels concerns the surrounding environment: weather conditions and type of terrain vary remarkably in the open, while they are rather uniform in a tunnel, thus allowing to reduce the number of cases to be examined. For the scope of this preliminary study, the simplified approach, with a single average weather condition, was used to assess the impact areas associated to the final outcomes occurring inside tunnels; in the open, reference was made to six different weather conditions and consequences analysis was carried out using the software Trace 8.b, as described in details elsewhere (Bubbico et al., 2004).
1202
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
2.1.2. People in the impact area On-route population density can be regarded to be practically the same for open air and tunnel sections: its value can be estimated based on traffic data for the specific route portion, assuming an average occupancy of road vehicles (or passenger trains), and taking into account that road traffic is queued in case of an accident (Bubbico et al., 2004). For accidents occurring inside a tunnel it can be assumed that there are no additional people at risk, while, in the open, resident population may be affected. In this case, provided that census data are available, the resident population density may be taken into account, adopting average values within some typical impact distances from the route (Bubbico et al., 2004). 3. Study case The present study case consists in the transport of three typical hazardous materials (gasoline, LPG, liquefied chlorine) and of an inert gas (liquefied nitrogen) from the city of Genoa to the French border at Ventimiglia, in Northern Italy, by rail and road (see Figs. 1 and 2, respectively). The rail route includes 16 tunnels exceeding the length of 500 m, for a total of 29.6 km, which is about 20% of the total length of the route (145 km); similarly, the road route includes 28 tunnels exceeding the length of 500 m, for a total of 39.8 km, representing about 25% of the total length of the route (159 km). 3.1. Tunnel sections modelling While an established procedure (Bubbico et al., 2004) has been adopted for the open air sections of the itineraries, specific modifications and assumptions have been introduced with reference to the tunnels. The main characteristics, and the geometry, of road and rail tunnels were based on the actual characteristics of some existing facilities; for rail tunnels the maximum width was set at 9 m, with the ceiling of the tunnel 5.225 m above the tracks; similarly, for road tunnels (two lanes), a width of 9.5 m and a height of 6.5 m was set. Two accidental scenarios were assumed for each examined transportation case, i.e. a partial release (15 mm hole, 15 min duration) and the catastrophic release of the whole tanker content through a 220-mm hole. The amount of products transferred in each rail and road tanker, and potentially involved in the release scenario, was assumed as 60 and 30 m3 , respectively. Specific event trees were developed to study the scenario evolution inside the tunnels for each transported product, which allowed to estimate the frequency of their respective final outcome cases, for both the assumed release scenarios (Amoroso, 2005; Parruccini, 2005). Then, the simplified consequence analysis was carried out, according to PIARC quantitative risk assessment models (OECD, 2003) to estimate the impact areas for each accidental scenario and transported product. The rather conservative hypothesis of no mechanical ventilation was assumed for consequence analysis calculations. In the case of fire, the toxicity of smoke was taken into account, in terms of CO production and dispersion (OECD, 2003). In the case of release of nitrogen, the potential asphyxiating effect due to the reduction of oxygen concentration was accounted for, assuming the following probit equation (Amoroso, 2005; Parruccini, 2005): 2
Y = −1582.64 + 55.02 ln (c t)
(1)
where Y is the probit, c is the nitrogen concentration (ppm) and t is the time (min). In all dispersion scenarios, probit equations were used to estimate exposure times corresponding to the death of the exposed
population (Y = 5). Then, the time was converted into distance, conservatively assuming a 5-min delay before involved people realise the need for escaping from the tunnel, and an average evacuation speed of 1.2 m/s. For thermal radiation and overpressure effects, the impact distances were derived from probit equations, based on the trends of these variables with the distance. The average temperature inside the tunnel was assumed to remain constant at 10 ◦ C (the route runs in Northern Italy). The people at risk was conservatively estimated assuming that dispersion occurs in the direction where people is more numerous, and that the accident takes place in the middle of each tunnel. In case of a rail accident, it was assumed that a derailment occurs, and that a passenger train is simultaneously blocked in the tunnel. The passenger train is assumed to consist of 12 cars, each 25 m long, carrying a total of 900 persons, giving rise to an average population density of 3 people/m of tunnel. However, the probability that a passenger train will be running in the tunnel exactly while an accident is occurring to the freight train is very low. In case of a road accident, the traffic is assumed to be stopped (on both the two lanes) upstream of the accident, while the vehicles will escape freely downstream of it. It is also assumed that the “average” road vehicle will carry two people and take 8 m of tunnel length, giving rise to an average population density of 0.5 people/m of tunnel. In this case, due to the traffic on the road, it is highly probable that a great number of vehicles will be present in the tunnel when the accident occurs. The details about tunnel sections modelling and calculations are reported elsewhere (Amoroso, 2005; Parruccini, 2005). 3.2. Risk analysis calculations 100 trips/year were assumed for all the transportation cases under exam, the scope of the simulation being that of comparing routes including and not including tunnels, and not that of actually assessing the risk associated to specific transportation cases. Risk analysis was carried out using TrHazGis software for risk assessment and management (Bubbico et al., 2004). The tool performs societal risk calculations based on: • A GIS (Geographical Information System) database, containing the detailed Italian rail and road networks, divided into portions few hundred meters long. To each route portion, the information concerning local accident rate, seasonal average weather conditions, average resident population density within a distance of 150 and 1500 m on the route, and on-route population, is associated. • A hazardous materials database, containing, for each listed product, and for both road and rail transport: - the probabilities of 2 release scenarios (partial release and catastrophic release), following an accident; - the probability of the evolutions of the assumed release scenarios into their possible final outcomes (different types of fires and/or explosion and/or toxic cloud); - the impact areas and distances for each outcome case, estimated by applying consequence analysis modelling under six different weather conditions. The software allows to automatically select the route based on its origin and destination, and performs transportation risk analysis calculations by combining route-related and product-related information, to obtain the relevant F–N curve. It has to be remarked that, in its standard use, the tool does not distinguish among route portions in the open and in tunnels and, indeed, it assumes tunnel sections as if they were open sections.
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
1203
Fig. 1. Rail route of the study case.
This assumption would provide incorrect values of the frequency of the accidental scenarios, of impact distances and of the population at risk, for the sections inside tunnels. Consequently, the tool was modified in order to take into account correctly the tunnel sections included in the road and rail routes under exam. 4. Results The study cases were first simulated using the software in its standard version, i.e. assuming each route as running fully in the open. The obtained results are shown in Fig. 3 (thin curves). Then, the study cases simulations were repeated taking into account the tunnel sections, i.e. substituting the reference data used in the open with those specific for tunnels. The results obtained
from these correct simulations are compared to the previous ones in Fig. 3, where they are shown as thick curves. 4.1. Rail transport By observing the societal risk curves shown in Fig. 3a–c, it can be noticed that, for the route under exam, the presence of the tunnels causes the societal risk to decrease, with respect to the case of travelling the whole route in the open. In fact, in the high frequency–low fatality zone and for all the examined hazardous products (chlorine, gasoline, LPG), the F–N curves including tunnels lay (even if very slightly) below those obtained fully in the open, while they are practically the same in the low frequency–high fatality zone. This result mainly depends on the lower number of people at risk
Fig. 2. Road route of the study case.
1204
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
Fig. 3. F–N curves for: (a) rail transportation of chlorine; (b) rail transportation of gasoline; (c) rail transportation of LPG; (d) rail transportation of nitrogen; (e) road transportation of chlorine; (f) road transportation of gasoline; (g) road transportation of LPG; and (h) road transportation of nitrogen.
inside tunnels with respect to that in the open. In fact, the probability of the simultaneous presence of a passenger and a freight train in the tunnel is rather low and a large number of fatalities is expected only in the open sections of the route, where the resident population density is high. Furthermore, the observed risk reduction after tunnels are included, is more evident for chlorine and LPG than for gasoline, according to the larger impact areas of the
final outcomes for the first two products with respect to the last one. 4.2. Road transport By observing the societal risk curves shown in Fig. 3e–g, it can be noticed that the risk generally increases when the route includes
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
tunnels. This is especially true for chlorine and gasoline, while, for LPG, in the high frequency–low fatality zone, the route including tunnels appears safer than the other. It has to be remarked that, for road transport, the population at risk inside tunnels may be lower as well as higher than that in the open section, depending on the particular route portion under exam. In fact, most of the route runs rather distant from towns and villages, giving rise to low resident population density, while, due to heavy traffic, the number of vehicles possibly blocked inside the tunnel in case of accident may be relevant. However, the different results obtained for the investigated products mainly depend on the different evolution of the accidental scenarios, in the open and inside tunnels, according to their respective event tree. This aspect involves the probability associated to the final outcomes and, especially, those associated to large impact distances. From this point of view, differences are limited when chlorine and gasoline are concerned: for chlorine the outcome is in both cases a toxic cloud with very high impact distances; for gasoline, inside a tunnel, a VCE is more probable than a flash fire, but the respective impact distances are of the same order of magnitude. On the contrary, large differences occur for LPG: for this product, a release in the open, may originate a variety of hazardous and less hazardous outcomes (for example, a jet fire), while, inside a tunnel, the generation of more hazardous outcomes (BLEVE, fireball, VCE) possibly causing a larger number of fatalities, is much more likely.
1205
rying hazardous materials through rail and road tunnels. For the rail transportation study cases, the societal risk appears (slightly) lower for the route including tunnels rather than for that running in the open. Risk analysis calculations carried out without considering the presence of tunnels appear to give rise to conservative results, and a specific analysis of tunnel sections may be omitted. Based on these results it seems unnecessary to impose special restrictions to the circulation of dangerous goods through rail tunnels. On the contrary, for road transport, the societal risk for itineraries including tunnels is generally higher than that for routes running completely in the open: in this case, according to the present results, the opportunity of allowing/forbidding the circulation of some hazardous goods through road tunnels, should be assessed for each specific case adopting a risk analysis approach. It has to be stressed that, even if these results were obtained with reference to a study case in Italy, the conclusions can be generalized to all cases where the travelled route passes by rather populated areas. Acknowledgments The contribution of F. Amoroso and M. Parruccini to the computational work is gratefully acknowledged. References
4.3. Transport of nitrogen The transport of nitrogen is a particular case, since the release of this inert product in the open, where it will quickly disperse, would not actually represent a risk issue. On the contrary, inside a tunnel, due to poor dispersion, the sudden spill of a large amount of liquefied nitrogen may give rise to asphyxiating effects. By examining the F–N curves relevant to the routes including tunnels, shown in Fig. 3d and h, it can be noticed that, for rail transport, the risk is extremely low, even if the potential number of fatalities is rather high, while, for road transport, the risk is not negligible, even if the potential number of fatalities is not particularly high. However, in both cases it seems unnecessary to forbid the circulation of this product through road or rail tunnels. 5. Conclusion From the test cases examined in the present work, some important preliminary considerations can be drawn about the opportunity to allow, forbid or limit the circulation of vehicles car-
Amoroso, F., 2005. Development of a Methodology for Risk Assessment of Hazardous Goods Transport Through Road Tunnels (in Italian). Universita` di Roma “La Sapienza”. Bubbico, R., Di Cave, S., Mazzarotta, B., 2004. Risk analysis for road and rail transport of hazardous materials: a GIS approach. J. Loss Prev. Proc. Ind. 17 (6), 483–488. CCPS, 1995. Guidelines for Chemical Transportation Risk Analysis. AIChE, New York. CCPS, 2000. Guidelines for Chemical Process Quantitative Risk Analysis, 2nd ed. AIChE, New York. CD, 1994. Council directive 94/55/EC. Off. J. Eur. Commun. L 319, Bruxelles, 12.12.1994. CD, 1996. Council directive 96/49/EC. Off. J. Eur. Commun. L 235, Bruxelles, 17.9.1996. CD, 2004. Council directive 2004/54/EC. Off. J. Eur. Commun. L 167, Bruxelles, 30.4.2004. MIT, 2005. Ministero delle Infrastrutture e dei Trasporti, Safety in rail tunnels (in Italian). Gazzetta Ufficiale, n.89. OECD/PIARC, ERS2, 1997. Transport of Dangerous Goods Through Road Tunnels. Technical Reports on Current National and International Regulations. OECD/PIARC, ERS2, 2003. Transport of Dangerous Goods Through Road Tunnels. Quantitative Risk Assessment Model version 3.60 Reference Manual. Parruccini, M., 2005. Development of a Methodology for Risk Assessment of Hazardous Goods Transport Through Rail Tunnels (in Italian). Universita` di Roma “La Sapienza”. United Nations, 2003. Recommendation of the multidisciplinary group of experts on safety in tunnels (rail). TRANS/AC. 9/9, 1.12.2003.
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Preliminary study on the transport of hazardous materials through tunnels Roberto Bubbico, Sergio Di Cave, Barbara Mazzarotta ∗ , Barbara Silvetti Dipartimento di Ingegneria Chimica Materiali Ambiente, Universit` a di Roma “La Sapienza”, Via Eudossiana 18, 00184 Rome, Italy
a r t i c l e
i n f o
Article history: Received 25 September 2007 Received in revised form 4 April 2008 Accepted 25 May 2008 Keywords: Tunnel Transport Hazardous materials Risk analysis
a b s t r a c t The risk associated to road and rail transportation of some hazardous materials along two routes, one including a significant portion in tunnels, and the other following the same path, but running completely in the open, is assessed. The results show that, for rail transport, no particular risk increase or mitigation is associated to the circulation of the dangerous goods through tunnels; on the contrary, for road transport, a risk increase is generally observed in the presence of tunnels. However, for LPG, the risk curve in the open lies above that in tunnels in the high frequency–low fatality zone, according to the different evolution of the accidental scenarios in the tunnel (assuming no ventilation). The transportation of liquefied nitrogen, not hazardous in the open but potentially asphyxiating in a tunnel, gives rise to a negligible risk when performed by rail, but to a not negligible one, when performed by road. These preliminary results focused on the risk for the exposed population, suggest that it may be unnecessary to limit dangerous goods circulation through rail tunnels, while, at least for some types of dangerous goods, the circulation through road tunnels may be allowed/forbidden based on the results of a specific risk analysis. © 2008 Elsevier Ltd. All rights reserved.
1. Introduction Long distance road and rail connections in Europe often include the passage through tunnels belonging to the National networks or connecting different Countries. Both these situations occur in Italy, with the long mountain chain of Apennines running from the north–west to the south–east of the Country, and with the Alps as boundary separating Italy from France, Switzerland, Austria and Slovenia. As a matter of fact, in Italy 83 road tunnels and 255 rail tunnels exceed 1 km in length; presently, three rail and three road tunnels longer than 10 km exist through Alps, with two more planned, whose length will exceed 50 km. A number of severe tunnel accidents occurred in the last years in Europe, with a high toll in terms of human lives, as well as direct costs for repairing the infrastructure and indirect costs associated with the temporary closure of the route including the gallery. It has to be remarked that, indeed, hazardous materials were not involved in any of these accidents; nevertheless, even common freight and passenger vehicles demonstrated their potential for generating a hazardous accidental scenario inside a tunnel, where the confinement causes longer persistence and higher levels of temperature, of concentration of toxic gases, and of overpressure with respect to
∗ Corresponding author. Tel.: +39 06 4458590; fax: +39 06 4827453. E-mail addresses: [email protected] (R. Bubbico), [email protected] (S. Di Cave), [email protected] (B. Mazzarotta), [email protected] (B. Silvetti). 0001-4575/$ – see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.aap.2008.05.011
open air. Moreover, people inside a tunnel may experience difficulties when trying to escape due to poor visibility, distant location of safe places and difficulty to locate them, and stressing environmental conditions. The need for increasing tunnel safety requirements is well present both to Public Authorities and to the general public, and some specific provisions are being issued for long rail and road tunnels. 1.1. Rail tunnels A group of experts from the United Nations Inland Transport Committee (United Nations, 2003) has studied the safety of tunnels in the range 1–15 km (for tunnel exceeding 15 km specific additional safety measures are enforced). The outline of the guidelines is that of requiring a basic, simplified, risk analysis for tunnels ranging from 2 to 9 km, and a standard risk analysis for tunnels exceeding 9 km. A regulation issued by the Italian Ministry of Infrastructures and Transportation (MIT, 2005), concerning safety requirements for rail tunnels, considers all the aspects involving infrastructure, exercise, users, and carriages. The requirements are listed according to the following objectives: • Incident prevention; • Consequence mitigation;
1200
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
• Evacuation of the passengers; • Emergency relief. A set of minimum requirements applies to all rail tunnels, and no risk analysis is needed for tunnels where: • length does not exceed 2 km; • railway traffic is less than 220 trains/day; • trains carrying hazardous materials and passenger trains cannot simultaneously circulate in the tunnel; • there is no slope inversion in the tunnel; • there are no specific area risks close to the tunnel inlets. Risk analysis has to be performed for all tunnels not fulfilling the previous requirements, and additional provisions may be set to grant a sufficient safety level. The risk analysis procedure is specified, and the basic scenario is a fire in a car of a passenger train. Acceptability thresholds are set for both individual and societal risk: for example, the ALARP limits are 1 × 10−9 –1 × 10−11 events per passenger and per km for individual risk. 1.2. Road tunnels The European Directive 2004/54/EC (CD, 2004) sets the minimum safety requirements of road tunnels exceeding 500 m, according to a two-steps approach: 1. a minimum safety level will be granted to all road tunnel users; 2. the sustainability of the present transportation system, especially in mountainous districts, will be verified, possibly adopting alternative transportation modes and routes. Accident prevention and mitigation of the consequences are obtained by adopting proper management and technical standards. The organizing aspects define jurisdiction and tasks of some key figures: • Administrative authorities, with the general responsibility of tunnel safety at a national, regional or local level. • Inspection services, assessing tunnel safety by means of tests and inspections. • Tunnel manager, responsible for tunnel exercise. • Safety managers, responsible for implementing all tunnel prevention and safety measures. Technical requirements concern: • Infrastructures, classified in five levels, depending on tunnel type, volume, traffic and length. To each level specific features of the tunnel (number of pipes, escape routes, ventilation, emergency exits, and possibly, special provisions) should correspond. A minimum set of requirements concerning escape routes, extinguishers, radio broadcasting, video-surveillance, cabling specific road signals to indicate escape routes and safe place location will apply to practically all tunnels. • Rules concerning works in tunnels, incidents management, activities of the control center (when present), tunnel closure, hazardous materials transportation, vehicle passing and minimum vehicle distance. • Vehicles, with provisions concerning the presence of extinguishers on the vehicles, the prohibition of filling additional fuel tanks and, possibly, additional safety measures. • Information to the users, issuing regular informative campaigns about tunnel safety, instructing the public about proper behavior in case of incidents and the location of safety devices.
Most existing tunnels have been designed and constructed some decades ago, and it is well recognized that the costs for their adaptation to these more demanding standards may be extremely high: for this reason a 10 years deadline has been set, which may be extended 5 years further in the Countries with a particularly high presence of tunnels. 1.3. Dangerous goods circulation in tunnels Generally speaking, road and rail transportation of hazardous materials is ruled by ADR and RID codes, respectively (CD, 1994, 1996); however, these documents do not provide any indications related to the passage of vehicles carrying dangerous goods through tunnels. The guidelines cited in Section 1.1 (United Nations, 2003) suggest that, based on the results of risk analysis, it may be also possible to prohibit the simultaneous passage of hazardous goods and passenger trains in tunnel longer than 2 km. Similarly, the regulation issued by the Italian Ministry of Infrastructures and Transportation (MIT, 2005), as far as dangerous goods transportation is concerned, relies only on operating procedures for preventing incidents. In particular, trains carrying hazardous materials and passenger trains will not be allowed to simultaneously circulate in the tunnel, which will be obtained by adopting a proper scheduling of the trips or limiting the transit to dangerous goods. This lack of general rules persists also in the case of road tunnels. As a matter of fact, a study concerning road tunnels (OECD, 1997) showed that no general code exists, and also in the European Directive previously cited in Section 1.2 (CD, 2004), no specific reference is made to hazardous materials circulation in tunnels, even if the general measures suggested may be useful also in such a case. Presently, however, regulations concerning hazardous materials circulation are issued only for specific (long) tunnels and, especially, for those connecting different Countries: the circulation of hazardous materials in the tunnel may be free, limited to specific ADR/RID products, or forbidden. By examining the actual situation for some long European tunnels, it seems that no particular restrictions apply to rail circulation of RID materials (with the only exception of the Eurotunnel below the English Channel), while, for the road case, the circulation may be prohibited to some classes of ADR materials (tunnel of Frejus) or to all of them (tunnel of Mont Blanc). The issue concerning the circulation of dangerous products is well present also from the planning stage of two new European tunnels (56 km between Innsbruck and Brenner, and 54 km on the new rail line connecting Lyon and Turin). It is clear that the subject concerning the circulation of hazardous materials in tunnels is rather complex and that any sound approach to the problem should be based on risk analysis. In fact, just forbidding dangerous goods circulation, from one hand will increase the safety of tunnel users, but, on the other hand, will submit to additional risks the population living close to or travelling along the alternative routes. In fact, it should be born in mind that alternative paths avoiding tunnels are generally much longer than direct routes, may present higher accident rates, and, above all, may cross a number of urban sites where the consequences of an accident may affect a greater number of people with respect to tunnel users. In this paper, a transportation risk analysis (TRA) approach is applied to assess the risk associated to the transport of some different hazardous materials by road and by rail along routes including significant portions inside tunnels. The results are compared with those obtained assuming the same routes to run completely in open air, in order to preliminarily assess the extent of risk increase/decrease to be expected for each product and transport modality.
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
2. Procedure Transportation risk analysis (TRA) concerning hazardous materials is basically an extension of the well-known (CCPS, 2000) quantitative risk analysis (QRA). In particular, it takes into account that the risk source is not fixed, but travels in a continuously changing environment, by dividing the route into homogeneous portions where all the parameters involved in risk calculations (accident rate, scenario probability, population at risk, etc.) can be assumed to keep constant (CCPS, 1995). The risk is usually assessed in terms of societal risk, which requires to estimate: • the frequency of all the final outcomes deriving from the accidental scenarios; • the extension of the areas where the consequences of the final outcomes may cause fatalities to the exposed population; • the expected number of people in those impact areas. Accordingly, the information concerning the transport case (product amount and physical state, number of trips, seasonal/daily schedule of the trips) should be combined with those related to each route portion, and consequence analysis calculations have to be carried out to account for the different scenarios, outcome cases and environmental conditions. Specific software tools, often based on a GIS (Geographic Information System) to manage route-related information, have been developed to assist TRA (Bubbico et al., 2004). 2.1. Open air and tunnel sections A number of differences may be encountered when applying TRA to open air and tunnel portions of a route: • Accident rate: the different environmental conditions experienced inside a tunnel (for example, poor visibility, but also the absence of snow, fog and rain), as well as the possible enforcement of different speed limits or safety distances, may affect the accident rate. Accordingly, for road transport, the expected accident rate may increase, as well as decrease, with respect to an open air section; on the contrary, for rail transport, the accident rate is not expected to vary significantly in tunnels. • Accidental scenarios and final outcomes: the presence of a confinement inside a tunnel causes the accidental scenarios to deeply change. As a limit case, the release of liquefied nitrogen, which is an inert product, can be considered: in open air the spill will evaporate and disperse safely; in a tunnel, on the contrary, the sudden evaporation of a large amount of nitrogen may reduce local oxygen concentration, and possibly cause the death of the exposed population. • Given an accidental scenario, the range of possible final outcomes also changes: for example, the release of a flammable material in a tunnel, in case of ignition, besides thermal radiation (which is practically the only effect in the open) can generate a toxic, and hot, smoke cloud rapidly invading the tunnel. • Finally, the probability associated to each branch of the event tree, describing the evolution of the accidental scenario into the various final outcomes, may be different in the open and inside tunnels. For example, due to the confinement, a VCE is more likely to occur in a tunnel rather than in the open. • Consequences of the final outcomes: the extension of the impact area changes remarkably in a tunnel with respect to open air. Thermal radiation levels are higher, due to confinement, and more persistent, due to the thermal insulation provided by the tunnel structure. Peak overpressures increase, too, due to the (partial)
1201
reflection of the shock waves operated by tunnel walls and vault. Toxic substances concentrations reach higher values, and the toxic clouds may easily pervade the whole section of the tunnel, in case of both buoyant and dense clouds, and for an extension of many hundreds meters. • People in the impact area: in the open both residents and on-route population can be affected by the consequences of the release of a hazardous material, while, in a tunnel, only on-route population is at risk. However, in the open, people can easily escape from the accident location, which is not true in a tunnel. For the scope of this preliminary study, the differences possibly occurring in the accident rates between open air and tunnel sections of the route have been disregarded, and the attention has been focused on accidental scenarios, and on outcome cases probabilities and consequences. 2.1.1. Accidental scenarios and final outcomes The accidental scenarios for hazardous materials transportation usually consist in the spill of the product from holes of different sizes, ranging from a puncture to a catastrophic rupture capable of rapidly emptying the container. Generally, TRA in the open takes into account the release of flammable and toxic liquid or gases; in the case of tunnels, in addition to these products, inert gases (both gaseous and liquefied) have to be considered, since their release may result asphyxiating in a confined environment. The final outcomes assumed in the open depend on the physical state and on the characteristics of the hazardous material: • • • •
Flammable liquid: pool fire, flash fire, VCE. Flammable gas: jet fire, flash fire, VCE. Flammable liquefied gas: jet fire, flash fire, VCE, BLEVE, fireball. Toxic gas or liquid: toxic cloud.
For a tunnel, due to the containment, the final outcomes are assumed as follows (OECD, 2003): • Flammable liquid: pool fire, flash fire, VCE, toxic cloud. • Flammable gas: jet fire, flash fire, VCE, toxic cloud. • Flammable liquefied gas: jet fire, flash fire, VCE, BLEVE, fireball, toxic cloud. • Toxic or inert gas or liquid: toxic cloud. The models to be used for determining the heat radiation, overpressure and concentration profiles in the open and inside a tunnel are also different. In the former case a number of mathematical models are reported in the literature (CCPS, 2000) and consequence analysis software are available; in the latter case, the proposed approaches may be based either on very simplified models, like those proposed by the international group of experts working on tunnel safety (OECD, 2003) or on very demanding mathematical modelling, performed by means of Computational Fluid Dynamics (CFD). Another difference between open air and tunnels concerns the surrounding environment: weather conditions and type of terrain vary remarkably in the open, while they are rather uniform in a tunnel, thus allowing to reduce the number of cases to be examined. For the scope of this preliminary study, the simplified approach, with a single average weather condition, was used to assess the impact areas associated to the final outcomes occurring inside tunnels; in the open, reference was made to six different weather conditions and consequences analysis was carried out using the software Trace 8.b, as described in details elsewhere (Bubbico et al., 2004).
1202
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
2.1.2. People in the impact area On-route population density can be regarded to be practically the same for open air and tunnel sections: its value can be estimated based on traffic data for the specific route portion, assuming an average occupancy of road vehicles (or passenger trains), and taking into account that road traffic is queued in case of an accident (Bubbico et al., 2004). For accidents occurring inside a tunnel it can be assumed that there are no additional people at risk, while, in the open, resident population may be affected. In this case, provided that census data are available, the resident population density may be taken into account, adopting average values within some typical impact distances from the route (Bubbico et al., 2004). 3. Study case The present study case consists in the transport of three typical hazardous materials (gasoline, LPG, liquefied chlorine) and of an inert gas (liquefied nitrogen) from the city of Genoa to the French border at Ventimiglia, in Northern Italy, by rail and road (see Figs. 1 and 2, respectively). The rail route includes 16 tunnels exceeding the length of 500 m, for a total of 29.6 km, which is about 20% of the total length of the route (145 km); similarly, the road route includes 28 tunnels exceeding the length of 500 m, for a total of 39.8 km, representing about 25% of the total length of the route (159 km). 3.1. Tunnel sections modelling While an established procedure (Bubbico et al., 2004) has been adopted for the open air sections of the itineraries, specific modifications and assumptions have been introduced with reference to the tunnels. The main characteristics, and the geometry, of road and rail tunnels were based on the actual characteristics of some existing facilities; for rail tunnels the maximum width was set at 9 m, with the ceiling of the tunnel 5.225 m above the tracks; similarly, for road tunnels (two lanes), a width of 9.5 m and a height of 6.5 m was set. Two accidental scenarios were assumed for each examined transportation case, i.e. a partial release (15 mm hole, 15 min duration) and the catastrophic release of the whole tanker content through a 220-mm hole. The amount of products transferred in each rail and road tanker, and potentially involved in the release scenario, was assumed as 60 and 30 m3 , respectively. Specific event trees were developed to study the scenario evolution inside the tunnels for each transported product, which allowed to estimate the frequency of their respective final outcome cases, for both the assumed release scenarios (Amoroso, 2005; Parruccini, 2005). Then, the simplified consequence analysis was carried out, according to PIARC quantitative risk assessment models (OECD, 2003) to estimate the impact areas for each accidental scenario and transported product. The rather conservative hypothesis of no mechanical ventilation was assumed for consequence analysis calculations. In the case of fire, the toxicity of smoke was taken into account, in terms of CO production and dispersion (OECD, 2003). In the case of release of nitrogen, the potential asphyxiating effect due to the reduction of oxygen concentration was accounted for, assuming the following probit equation (Amoroso, 2005; Parruccini, 2005): 2
Y = −1582.64 + 55.02 ln (c t)
(1)
where Y is the probit, c is the nitrogen concentration (ppm) and t is the time (min). In all dispersion scenarios, probit equations were used to estimate exposure times corresponding to the death of the exposed
population (Y = 5). Then, the time was converted into distance, conservatively assuming a 5-min delay before involved people realise the need for escaping from the tunnel, and an average evacuation speed of 1.2 m/s. For thermal radiation and overpressure effects, the impact distances were derived from probit equations, based on the trends of these variables with the distance. The average temperature inside the tunnel was assumed to remain constant at 10 ◦ C (the route runs in Northern Italy). The people at risk was conservatively estimated assuming that dispersion occurs in the direction where people is more numerous, and that the accident takes place in the middle of each tunnel. In case of a rail accident, it was assumed that a derailment occurs, and that a passenger train is simultaneously blocked in the tunnel. The passenger train is assumed to consist of 12 cars, each 25 m long, carrying a total of 900 persons, giving rise to an average population density of 3 people/m of tunnel. However, the probability that a passenger train will be running in the tunnel exactly while an accident is occurring to the freight train is very low. In case of a road accident, the traffic is assumed to be stopped (on both the two lanes) upstream of the accident, while the vehicles will escape freely downstream of it. It is also assumed that the “average” road vehicle will carry two people and take 8 m of tunnel length, giving rise to an average population density of 0.5 people/m of tunnel. In this case, due to the traffic on the road, it is highly probable that a great number of vehicles will be present in the tunnel when the accident occurs. The details about tunnel sections modelling and calculations are reported elsewhere (Amoroso, 2005; Parruccini, 2005). 3.2. Risk analysis calculations 100 trips/year were assumed for all the transportation cases under exam, the scope of the simulation being that of comparing routes including and not including tunnels, and not that of actually assessing the risk associated to specific transportation cases. Risk analysis was carried out using TrHazGis software for risk assessment and management (Bubbico et al., 2004). The tool performs societal risk calculations based on: • A GIS (Geographical Information System) database, containing the detailed Italian rail and road networks, divided into portions few hundred meters long. To each route portion, the information concerning local accident rate, seasonal average weather conditions, average resident population density within a distance of 150 and 1500 m on the route, and on-route population, is associated. • A hazardous materials database, containing, for each listed product, and for both road and rail transport: - the probabilities of 2 release scenarios (partial release and catastrophic release), following an accident; - the probability of the evolutions of the assumed release scenarios into their possible final outcomes (different types of fires and/or explosion and/or toxic cloud); - the impact areas and distances for each outcome case, estimated by applying consequence analysis modelling under six different weather conditions. The software allows to automatically select the route based on its origin and destination, and performs transportation risk analysis calculations by combining route-related and product-related information, to obtain the relevant F–N curve. It has to be remarked that, in its standard use, the tool does not distinguish among route portions in the open and in tunnels and, indeed, it assumes tunnel sections as if they were open sections.
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
1203
Fig. 1. Rail route of the study case.
This assumption would provide incorrect values of the frequency of the accidental scenarios, of impact distances and of the population at risk, for the sections inside tunnels. Consequently, the tool was modified in order to take into account correctly the tunnel sections included in the road and rail routes under exam. 4. Results The study cases were first simulated using the software in its standard version, i.e. assuming each route as running fully in the open. The obtained results are shown in Fig. 3 (thin curves). Then, the study cases simulations were repeated taking into account the tunnel sections, i.e. substituting the reference data used in the open with those specific for tunnels. The results obtained
from these correct simulations are compared to the previous ones in Fig. 3, where they are shown as thick curves. 4.1. Rail transport By observing the societal risk curves shown in Fig. 3a–c, it can be noticed that, for the route under exam, the presence of the tunnels causes the societal risk to decrease, with respect to the case of travelling the whole route in the open. In fact, in the high frequency–low fatality zone and for all the examined hazardous products (chlorine, gasoline, LPG), the F–N curves including tunnels lay (even if very slightly) below those obtained fully in the open, while they are practically the same in the low frequency–high fatality zone. This result mainly depends on the lower number of people at risk
Fig. 2. Road route of the study case.
1204
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
Fig. 3. F–N curves for: (a) rail transportation of chlorine; (b) rail transportation of gasoline; (c) rail transportation of LPG; (d) rail transportation of nitrogen; (e) road transportation of chlorine; (f) road transportation of gasoline; (g) road transportation of LPG; and (h) road transportation of nitrogen.
inside tunnels with respect to that in the open. In fact, the probability of the simultaneous presence of a passenger and a freight train in the tunnel is rather low and a large number of fatalities is expected only in the open sections of the route, where the resident population density is high. Furthermore, the observed risk reduction after tunnels are included, is more evident for chlorine and LPG than for gasoline, according to the larger impact areas of the
final outcomes for the first two products with respect to the last one. 4.2. Road transport By observing the societal risk curves shown in Fig. 3e–g, it can be noticed that the risk generally increases when the route includes
R. Bubbico et al. / Accident Analysis and Prevention 41 (2009) 1199–1205
tunnels. This is especially true for chlorine and gasoline, while, for LPG, in the high frequency–low fatality zone, the route including tunnels appears safer than the other. It has to be remarked that, for road transport, the population at risk inside tunnels may be lower as well as higher than that in the open section, depending on the particular route portion under exam. In fact, most of the route runs rather distant from towns and villages, giving rise to low resident population density, while, due to heavy traffic, the number of vehicles possibly blocked inside the tunnel in case of accident may be relevant. However, the different results obtained for the investigated products mainly depend on the different evolution of the accidental scenarios, in the open and inside tunnels, according to their respective event tree. This aspect involves the probability associated to the final outcomes and, especially, those associated to large impact distances. From this point of view, differences are limited when chlorine and gasoline are concerned: for chlorine the outcome is in both cases a toxic cloud with very high impact distances; for gasoline, inside a tunnel, a VCE is more probable than a flash fire, but the respective impact distances are of the same order of magnitude. On the contrary, large differences occur for LPG: for this product, a release in the open, may originate a variety of hazardous and less hazardous outcomes (for example, a jet fire), while, inside a tunnel, the generation of more hazardous outcomes (BLEVE, fireball, VCE) possibly causing a larger number of fatalities, is much more likely.
1205
rying hazardous materials through rail and road tunnels. For the rail transportation study cases, the societal risk appears (slightly) lower for the route including tunnels rather than for that running in the open. Risk analysis calculations carried out without considering the presence of tunnels appear to give rise to conservative results, and a specific analysis of tunnel sections may be omitted. Based on these results it seems unnecessary to impose special restrictions to the circulation of dangerous goods through rail tunnels. On the contrary, for road transport, the societal risk for itineraries including tunnels is generally higher than that for routes running completely in the open: in this case, according to the present results, the opportunity of allowing/forbidding the circulation of some hazardous goods through road tunnels, should be assessed for each specific case adopting a risk analysis approach. It has to be stressed that, even if these results were obtained with reference to a study case in Italy, the conclusions can be generalized to all cases where the travelled route passes by rather populated areas. Acknowledgments The contribution of F. Amoroso and M. Parruccini to the computational work is gratefully acknowledged. References
4.3. Transport of nitrogen The transport of nitrogen is a particular case, since the release of this inert product in the open, where it will quickly disperse, would not actually represent a risk issue. On the contrary, inside a tunnel, due to poor dispersion, the sudden spill of a large amount of liquefied nitrogen may give rise to asphyxiating effects. By examining the F–N curves relevant to the routes including tunnels, shown in Fig. 3d and h, it can be noticed that, for rail transport, the risk is extremely low, even if the potential number of fatalities is rather high, while, for road transport, the risk is not negligible, even if the potential number of fatalities is not particularly high. However, in both cases it seems unnecessary to forbid the circulation of this product through road or rail tunnels. 5. Conclusion From the test cases examined in the present work, some important preliminary considerations can be drawn about the opportunity to allow, forbid or limit the circulation of vehicles car-
Amoroso, F., 2005. Development of a Methodology for Risk Assessment of Hazardous Goods Transport Through Road Tunnels (in Italian). Universita` di Roma “La Sapienza”. Bubbico, R., Di Cave, S., Mazzarotta, B., 2004. Risk analysis for road and rail transport of hazardous materials: a GIS approach. J. Loss Prev. Proc. Ind. 17 (6), 483–488. CCPS, 1995. Guidelines for Chemical Transportation Risk Analysis. AIChE, New York. CCPS, 2000. Guidelines for Chemical Process Quantitative Risk Analysis, 2nd ed. AIChE, New York. CD, 1994. Council directive 94/55/EC. Off. J. Eur. Commun. L 319, Bruxelles, 12.12.1994. CD, 1996. Council directive 96/49/EC. Off. J. Eur. Commun. L 235, Bruxelles, 17.9.1996. CD, 2004. Council directive 2004/54/EC. Off. J. Eur. Commun. L 167, Bruxelles, 30.4.2004. MIT, 2005. Ministero delle Infrastrutture e dei Trasporti, Safety in rail tunnels (in Italian). Gazzetta Ufficiale, n.89. OECD/PIARC, ERS2, 1997. Transport of Dangerous Goods Through Road Tunnels. Technical Reports on Current National and International Regulations. OECD/PIARC, ERS2, 2003. Transport of Dangerous Goods Through Road Tunnels. Quantitative Risk Assessment Model version 3.60 Reference Manual. Parruccini, M., 2005. Development of a Methodology for Risk Assessment of Hazardous Goods Transport Through Rail Tunnels (in Italian). Universita` di Roma “La Sapienza”. United Nations, 2003. Recommendation of the multidisciplinary group of experts on safety in tunnels (rail). TRANS/AC. 9/9, 1.12.2003.