Introduction

1The main goal of the research behind is to elaborate a common method to assess industrial risks simultanily around industrial plants and along the transport road axes where hazardous substances are transported. It focuses on the geography of a firm, giving attention to the tight articulation between site and flows, static and dynamic aspects of the industrial running.

2The context of the attention paid to this kind of problem is the following : major hazards are now considered a social and environmental problem, for natural as well as for technological hazards. They draw the attention of public opinion, as well as politics, scientific and industrial interest (especially in the northwestern countries) (CF EWALD François (1990)). For industrial risks I'm interested in, which form part of the so-called technological risks, we can say that this multiface attention has enwidened both under "objective" reasons and more "subjective" or social reasons. First, the dangers are really increasing, even if the security systems continue to be improved. More and more chemical products are manufactured and transported and the uncertainty about their effects on the health and environment are widely unknown. Meanwhile, territories, ground, space become every day more sensitive (or vulnerable) : towns are spreading, subjecting every day more people to danger. In the same time, the psycho-sociological context helps to increase the attention given to major hazards. On one hand, the impact of risks is enhanced by intensive media coverage made possible by communication advances. On the other hand, ecology, search of life quality and security as well as environmental topics develop the social demand for less nuisances such as pollution and risks.

3Under those effects, risk assessment and management has been developed for years, in order to prevent those risks. Various security policies already exist. Among them, urban regulatory measures are proposed, in order to limit land uses, and though, to limit the potential damages resulting of an industrial accident.

4In a first part, we'll see how important is the spatial dimension to define the risk level. This importance given to space explains the town planning measures proposed to reduce risks, especially in town areas. This kind of regulatory approach leads us to emphasise the advantages of spatial analysis for risk assessment.

5But, as technological hazards are not only those of industrial plants, but also the risks related to hazardous material transportation, in a second part, a method to compare those different kinds of industrial risks is developed. This method is implemented on a case study : the storage and distribution of fuel oils in the Ile-de-France region.

6Finally, the calculations and comparisons of risk help to point out that freight transportation has to be better considered while dealing with industrial politics and towns planning propositions. It becomes very clear that the whole enterprise geography must be observed to limit technological risks. Thus, in a last conclusive part, I will just mention paths to discussion, as it seems very useful to discuss the results and extend this spatial analysis to other cases and hypothesis.

The spatial approach of technological hazards

Definition of risks

8Risk = frequency of an event X gravity of its consequences

9The simultaneous variation of both terms of this equation, frequency and gravity, define the level of risks. Several categories can be schematically proposed by the following curve.

10In this definition, major hazard is characterised by very serious consequences, but a very rare occurrence.

Figure 1 : farmer curve

Source : Le risque majeur, Secrétariat d'Etat chargé de l'environnement et de la prévention des risques technologiques et naturels majeurs / Direction de l'eau et de la prévention des pollutions et des risques / Délégation aux risques majeurs.

11In order to reduce major hazards, we can act on one or/and the other term of the equation :

• To reduce the frequency of accidents, technic security actions can be led ; they are usually undertaken by entrepreneurial managers ;

• To prevent from too serious consequences, public policies plan to reduce the use of the surroundings of industrial plants in order them not to be too densely occupied.

12In this latter aim, we have two legislative corpuses that define land use policies :

• on one hand the so-called "législation des installations classées pour la protection de l'environnement" = Law concerning classified establishments for the protection of the environment (Loi n° 76-663 du 19 juillet 1976 relative aux installations classées pour la protection de l'environnement. - J. O. du 20 juillet 1976) ;

• on the other hand, special urbanistic rules. Since they were adopted in 1987 (Loi n° 87-565 du 22 juillet 1987), these rules force town planners to take into account the risks while proposing land use documents, from local (POS, plan d'occupation des sols = land use plan) to regional (Schéma directeur = regional master plan). Special non densified zones have to be left within the danger limits defined around an industrial plant.

13The two corpus mentionned join each other to restrain the land use in the surroundings of dangerous plants. The main objective of such measures is to reduce the human loss if a major accident were to occur. In France, an authoritative style has been chosen, with the deterministic approach. The "penalizing accident", from the worst possible danger, is the reference to define the limits of the almost "forbiden" zone.

14The European COMAH directive for Control of Major Accident Hazards (so-called “Seveso directive” in memory of the well-known accident occurred near Milano in 1976), reinforce the previous French dispositions, especially on the control of land use around dangerous installations.

15Such measures limit the urban expansion. It leads some town planners to propose to move dangerous plants out of the inner cities, where some of the old industries are still located nowadays. But in the meantime, they forgot to consider another aspect of industrial course : transportation. As industrial sites and the in-and-outflows provided by these sites are very closely linked, so are the risks of these two parts of industrial activity. When industrial geography is reorganised, it can possibly modify the level of risks around the plant, along the flows, or both.

16Tense flows "naturally" tend to increase goods transport. This idea of taking the dangerous industry away from urban area would probably reinforce this tendency. The question was, than, to evaluate the importance of risk transferral (from site to transports), if the whole geography of an enterprise were to be modified, and even if the flows would, of course, cross more rural areas than before.

17Such an approach underline the very close articulation of transport and storage/production of dangerous materials and, thus, the tightly dependant risks along an industrial channel ; it bases the assessment on the space-uses for industrial plants as well as for transport, as these uses are considered elements of prime importance to define the level of risks ; the areas submitted to risks are calculated following the planning laws, as a certain consensus exists to accept those areas.

18Than, it was necessary to propose a common method to estimate both aspects of the industrial running and be able to make comparison between site-risks (around the industrial dangerous plants) and transport-risks (along the trunk roads) in several potential locations for the plant. It seemed relevant to use spatial analyses to do so, as town planning is very concerned and because of the urban location of many industrial plants.

How to assess industrial risks with the help of a gis ?

19Elaboration of a method to measure risks related to plants (sites) and transport (flows) of hazardous materials. The example of fuel-oil storage and distribution through Ile-de-France.

20At the beginning of the decade, in the early 90', the Regional masterplan of Ile-de-France region was reviewed. Following the legal dispositions, the new masterplan was supposed to take major hazards into account. It was just after a real crisis on land values in the french capital. The need of ground was thus very high, and land planners imagine that it would be of some use to move some of the industrial plants away, to the limits of the region. The main petroleum storages were particularly aimed, as most of the 21 that we can count in the region have been built for quite a century : they are now near the inner zone of the built-up area (fig 2). This industrial channel is quite interesting to study because of the importance of petroleum distribution by road, along with a very dense common traffic flow.

Figure 2 – Location of the main petroleum storages in the region of Ile-de-France

Site-risks assessment

21The methodology is based on the technical approach defined for fixed industrial risks. It considers the danger area as it is calculated in danger studies and urban plans. Within this area, loss could be noted if a major accident occurs. This step follows insurantial processes to fix the taxes for the insured customers, as it was proposed for some environmental hazards (floods for example, CF Agence de l’eau Seine-Normandie et al. (1994), and SAGERI (1987) for example).

22We first define the danger area (fig 3), following the official ministerial instruction (Instruction technique ministérielle du 9 novembre 1989 relative aux dépôts aériens existants de liquides inflammables).

Figure 3 – The danger area around a petroleum storage

23As the aim is to count the potential damages in this area, a GIS (Geographic information system) was used. It was than possible to systemise the calculations, which was very useful for several reasons :

• To study various sites,

• To be able to repeat the calculation as soon as the simulations would need it, for new plants location, in the outer built-up area, and whenever the land uses change (or are planned to be modified),

• Most of all, it allowed difficult calculation on a regional scale, such as the transport aspect would require for the catchment area of the industrial plants (on a second step).

24The Land Use Map (Mode d'occupation des sols = MOS), a very precise cartography of the activities in Ile-de-France was used. It is worked out by IAURIF (Institut d'aménagement et d'urbanisme de la région Ile-de-France = Town planning Institute of the region of Ile-de-France) and regularly updated.

25The danger area was reported on a special layer of the regional GIS (fig 4). The damages were than counted : the area of each of the 130 activities included in this perimeter were isolated and unit damages assigned to these surfaces. The damages are expressed in lives lost on the one hand and economical loss on the other hand. The latter operation allows to sum the losses and propose synthetic indicators very usefull to the later comparison.

Figure 4 – The danger area as reported in the GIS of IAURIF, describing the land use

Transport-risks assessment

26The same operation was repeated for the flows of road distribution in the vast catchment zone. We first draw the distribution geography. The result is a cartography of the monthly flows to the customers of an oil storage, for the fuel-oil products used by service-stations, administration vehicles, freight and travelers transportation or public works enterprises of the whole region (fig 5).

Figure 5 – The monthly flows of petroleum from an inner-city oil storage to its customers

27Then we defined the corridor in which a tank-waggon accident would be felt (57 m large around the road axe = 33,5 feet radius) (fig 6). A synthetic indicator is elaborated. It correspond to a round area affected by a potential accident, and represents the medium accident. It somehow summarizes the catchment area configuration, which greatly depends of the areas crossed by the trucks (mostly urban if the storage is located in the inner built-up area, mostly rural on the contrary, if the storage is located far from this area).

Figure 6 – The corridor along road axes used by tank-waggon of oil during the distribution

Comparisons and simulations

28In order to help comparison, we use the annual chances of an accident to occur (based on statistics). It is much higher for a truck than for a storage plant, even if less serious for the former than for the latter. Two series of comparisons are made :

• between the storage-risks and the transport-risks

• in accordance to different locations of the storage facilities. We chose 3 test-sites from the center to boarder of the region, with attention payed to the previous density (as low as possible for the new storage) and to the in and out oil flows (possible connection with the pipeline, easy links with the road network) (fig7).

Figure 7 – New location of fuel oil plants for the simulation of flows and transport-risks

	actual location	simulated site 1	simulated site 2	simulated site 3
annual probability	0.0000144	0.0000144	0.0000144	0.0000144
total damages*	1 347 173 136	3 694 475	229 617 561	1 801 617
damages/year	19 413	37	3 309	26
expected death	94.87	17.51	106.33	1.98
annual probability of human loss	0.0013671	0.0002523	0.0015322	0.0000285

Table 1 : potential damages within the plant area

* unit : french francs (the financial expression allows comparisons)

	actual location	simulated site 1	simulated site 2	simulated site 3
annual probability	0.0104932	0.0105339	0.0121196	0.0439493
danger regional area (km2)	285 054 648	288 023 426	313 564 468	333 666 216
total damages*	439 282 434 814	432 621 778 739	458 660 639 129	474 811 125 577
damages for medium accident	15 721 539	15 337 087	14 935 747	14 530 179
damages for medium accident/year	164 969	161 559	181 015	638 591
expected death/medium accident	9.33	9.09	8.76	8.4
annual probability of human loss	0.0978989	0.0958001	0.1061119	0.3693832

Table 2 : potential damages within the merchant zone

* unit : french francs

29The results (tables 1 and 2) show that as the flows increase, the transport-risk also increase, but much more than the eventual gains of security around the plants. This can be noted even if :

• the new location allows to reduce site-risks. This is not an evident point, as the pollution cost may be higher in rural areas than in urban (meanwhile human losses may decrease),

• the areas crossed by the flows seem to be less sensitive (less traffic, less urban areas). Statistics reveal that the higher the speed, the bigger accident chances. So road tank-waggon accidents are much more frequent on the rural roads than on urban streets (luckily). We can, at the most, fear soil and water contaminations rather than death, which may be socialy more acceptable.

30This last remark leads me to the discussion of the results.

Conclusion : advantages and limits of such a method

31The general aim of our research was to emphasize that industrial risk along a whole industrial branch had to be closely considered when the public authorities try to reduce it. Considering only the industrial plants can lead to forget that transfer of risk is the only result of a new location for those plants. Our method helps to see simultaneously various sides of industrial risks, but, as many factors had to be taken into account, we are very conscient that it must be refined.

Necessary variations of the hypothesis,

32Our results are probabily very much indebted to the basic hypothesis : mode of calculation, determinist approach, normative references. It is a conscious choice which consider that a certain consensus exists among industrial enterprises and the french administration, about the definition of danger areas. But other parameters could be defined and would probabily allow the proposition of new scenarios of risks. Validity tests must be precisely done and margins of errors of the method be evaluated.

33It would than expand the use of such risk assessment which aimes to give documents and elements (maps, indexes, risk and social costs of dangerous activities) elaborated to discuss about the opportunity of industrial locations.

Other parameters to take in account

34Of course, the indicators, such as the amount of damages and the chances of accident to occur, are not the only parameters used to make urban or industrial plans and to lead environmental politics. Other parameters like employment rates, inner taxes of industry, social tolerance or acceptance of risk depending on the region and habits also play a role in those decisions. The previous propositions can only be a part of socio-economical debates.

Extension of the approach

35To improve the quantitative methodology previously exposed, it would be of interest to extend it in various ways and proposed other comparisons :

• On a spatial concerns, other regions and other scales could be explored. The basic problem is the one of the available data, which can scarcely be gathered because of industrial and administrative reluctancy, of unknown phenomenas and flows, etc. I'd like to better consider the vertical dimension than it can hardly be done with the planimetric approach chosen with the spatial analysis. The GIS allows all these extensions but adaptations must be done, depending on the system and the data available.

• Thematicaly, The application had to be limited in the present case because of the lack of data, but other industrial fields are also to be explored and other dangers considered. The intermodal comparison would be of interest too.