Introduction

1Dublin has experienced significant economic development since 1990 (Ellis & Kim, 2001), partly catching up with other European capitals, even though it appeared in the mid-1980s as a declining industrial city (McLaran, 1984, Crotty, 2000). This development poses a problem because the city of Dublin is confronted with recurrent floods (Lhomme, 2012). Over the past ten years, it has suffered three major floods (Leahy, 2011). In practice, the city of Dublin appears to be a "textbook case" for flood management in urban areas. Indeed, this territory has to cope with floods of various origins (river, coastal and rainwater floods) and with a continuous urban development which exerts an important land pressure on flood-prone areas.

2To cope with these floods, Dublin city managers claim to be implementing measures to improve the resilience of their territory (Leahy, 2011). However, since resilience is the subject of much debate in the academic field (Klein et al., 2003 ; Djament-Tran & Reghezza-Zitt, 2012 ; Lhomme & Djament-Tran, 2012), it is logical to question its operational scope (Reghezza-Zitt et al., 2012). Considered by some as a simple notion (Brand & Jax, 2007), by others as a concept (Dauphiné & Provitolo, 2007), even as a new paradigm (Hollnagel et al., 2006), resilience is also used in many disciplines that give multiple interpretations (Rufat, 2009 ; Lhomme et al., 2010). The objective of this article is to compare the "operational" use of resilience with its more "theoretical" use by using a concrete example.

3The implementation of measures to improve the resilience of this territory now seems imperative for managers and political actors. The presentation of the Dublin context will help to understand why. Then, the actions implemented by the managers of the city of Dublin, according to them relating to resilience, will be compared with academic acceptances in order to identify certain gaps. Finally, the resilience of Dublin's technical networks will be analysed to understand the resilience of this territory (Lhomme et al., 2010; Serre, 2011). It will then be necessary to determine whether the display of Dublin managers corresponds to what should be expected from a management policy centred on the concept of resilience.

Reconciling urban development and flood management

A "small" capital in full expansion

4One of the main characteristics of the Irish capital is its small population: with around 520,000 inhabitants (www.cso.ie) it is possible to consider Dublin as a "small" European capital. This observation must be qualified if we take into account the Dublin conurbation which counts a little more than one million inhabitants, or the Greater Dublin which gathers a little more than 1 800 000 inhabitants (www.cso.ie). Taking into account the neighbouring counties allows a better understanding of the reality of Dublin's territory, since the latter highlight another specificity of this territory: its low density.

5If the centre of Dublin is rather dense, about 4500 inhabitants/km², its urban area has on the contrary a very low density for a European capital, about 230 inhabitants/km² (www.urbanaudit.org). The density of Greater Dublin, which is 25 inhabitants/km², is a good illustration of the diversity of this area, comprising a fairly densely populated city centre, housing estate areas where the majority of the population is concentrated and predominantly rural areas in the hinterland.

6From an economic point of view, in the early 1980s, the Irish economy was mainly oriented towards agriculture (Crotty, 2000; Grison, 2004). In the late 1980s, Ireland moved towards a voluntary liberal (even neoliberal) policy of attracting foreign investors and concentrating the country's development on three sectors of activity: chemicals and pharmaceuticals, information and communication technologies, and financial services (Crotty, 2000). Logistics and financial assistance were provided for foreign companies and corporate income tax rates were lowered. Finally, a highly flexible labour market, combined with strong investment in education, provides Ireland with strong economic growth: 6% per annum between 1995 and 2007 (Grison, 2004). It is this development that has earned it the image of the "Celtic tiger" compared to the Asian tigers (Crotty, 1998).

7However, Irish growth has been concentrated mainly in the cities of Dublin and Cork, the main city in the south of the country (Crotty, 2000; Grison, 2004). This economic development is accompanied by a significant demographic increase (Fig. 1).

8Dublin is a small and growing capital. The population of Greater Dublin is expected to reach 2.4 million in 2020 (GDSDS, 2005). This development puts significant pressure on the city's overall technical infrastructure, such as transport (Ellis & Kim, 2001). Thus, the city of Dublin is experiencing significant traffic jams. According to rankings established by the GPS company TOMTOM and the road data producer NAVTEQ, Dublin would rank among the 30 most congested cities in Europe. Some months, the city of Dublin has even been ranked among the 10 busiest European cities (NAVTEQ, 2011).

9The 2008 financial crisis reexamines Ireland's economic development model: the city of Dublin is no longer ranked among the world's top 40 financial centres, even though it may have been among the top 10 in 2009. And yet Dublin's population still continues to grow (GDSDS, 2005). This demographic development calls into question a development policy based on increasing urban sprawl, while the necessary investments to resize the technical infrastructures are still awaited.

A city highly exposed to flooding

10On Monday 24 October 2011, heavy rain fell on the east coast of Ireland. Within three hours, rainfall reached 80 millimetres, equivalent to three days of precipitation (Maguire, 2012). Two rivers that flow through the city, the Dodder and Camac rivers, overflowed their banks, flooding about 300 homes and drowning two people (Leahy, 2011). At the same time, torrential rains flooded major roads and one of Dublin's largest shopping centres, the Merrion shopping centre. This major event required the initiation of an emergency plan (Leahy, 2011). He reminds that the city of Dublin is exposed to floods from various sources (Fig. 2).

11Dublin is a flood-prone city, primarily because of its geographical location (Fig. 3). Dublin is a coastal city (Irish Sea) crossed by three main rivers (Liffey, Dodder, Tolka), canals (Royal Canal, Grand Canal) and many other rivers (Wad, Poddle, Camac) some of which are now underground. In addition, it is subjected to very heavy rains and many storms. In addition, two rivers (the Liffey River and the Dodder River) meet in the heart of Dublin. This confluence, combined with the presence of numerous small rivers, creates a complex hydrological and hydraulic situation in the Liffey estuary.

12The city of Dublin is crossed and bounded by rivers whose catchment areas are rather small and reactive (Leahy, 2011), rain and river floods often add up, forming a major event. This also explains the attention given to forecasting and modelling these hazards by the Dublin City Council Water, Waste and Environment Department (Maguire, 2012). The potential effects of global warming on coastal flooding are of concern to CCD managers (Leahy, 2009). These managers thus mobilize significant resources to model this hazard and the effects of global warming on it. Stormwater flooding highlights the undersizing of the stormwater drainage system and its obsolescence in parts of Dublin City. Lessons learned from the October 2011 floods confirm that many sanitation network infrastructures need to be replaced and resized (Maguire, 2012). This undersizing is a concrete illustration of the pressure exerted by the development of the city of Dublin on its infrastructure.

Resilience: from theory to practice

13Constrained by the high exposure of its territory and by its urban development, the DCC has multiplied its actions in order to be able to cope with floods (Fig. 4). The February 2002 flood was the triggering event that led the DCC to review its management policy (Leahy, 2009).

Dublin and its resilience policy: managers' actions

14The DCC is currently based on the approach defined in the FloodResilienCity project, which is itself based on the Scottish 4A approach: Awareness, Allaviation, Avoidance, Assistance (Ashley et al., 2012). This approach is intended to be holistic: it encompasses the different phases of risk management and different measures, protection (with the construction of dykes), adaptation (with the construction of buildings considered resilient) and assimilation (acceptance of floods in certain areas to leave more space for water). The many actions undertaken by the DCC, which may at first seem disjointed, are now based on this general policy. This policy promotes measures to improve the resilience of their territory (Leahy, 2011), without the term resilience being defined or made explicit. It seems that for managers and politicians, all the actions undertaken since 2002 to deal with floods fall within this framework. What are the characteristics of these actions?

15This policy focuses on protective measures and the establishment of forecasting and warning systems (Leahy, 2009). These measures are now being implemented on the scale of the city's various catchment areas. For example, a detailed topographic and hydrological study of the Dodder River and its tributaries was carried out. The objective was to provide geolocalized data to model the functioning of this river. From this modelling, a precise identification of flood zones could be obtained. Forecast models coupled with an early warning system are now in the development phase (Leahy, 2011).

16Based on these models, a design office has developed specific measures to reduce the risks in the Dodder River catchment. Structural measures are designed to reduce hazards (construction of new dikes, dams, mobile barriers, etc.). Measures described as non-structural focus on spatial organisation and planning (such as the preservation of flood plains through specific zoning). This study proposes a comprehensive range of viable measures, taking into account a range of technical, economic and environmental criteria at the scale of the Dodder River watershed (Leahy, 2009). This approach was then applied for each watershed.

17In parallel with these measures, managers are trying to raise public awareness of flood risks in order to develop a certain risk culture (Leahy, 2011). Without totally breaking with its past policy based on protective measures, the DCC has therefore adapted and diversified its management methods. For example, the city is flooding the city: the growing waterproofing of soils and the undersizing of sanitation networks are leading them to overflow more and more frequently. The DCC, which is aware of these urban planning issues, is currently trying to deal with them by using local and technical solutions. The aim is to reduce runoff by greening urban areas, to allow floods to flood private gardens, and no longer to build dikes to protect oneself. At the urban scale, it is the development policies that are questioned. The general objective is to give more space to the water. The mitigation of urban runoff requires different modes of action and different legislation implementing preventive and curative measures.

18Contrary to what was proposed in the past, these strategies are no longer based on a principle of opposition to hazards (an "all-protection" strategy), but are intended to be more comprehensive and complete by providing complementary responses through the use of non-structural measures linked to urban planning and better spatial management. It is by combining on the one hand traditional flood control measures and on the other hand development actions beyond the framework of floods, that DCC managers claim to improve the resilience of their territory. For Dublin managers, resilience is based on the following triptych: an integrative and global approach (the 4As); non-structural measures combined with structural measures; the development of the risk culture among citizens (population, economic actors...).

19Nevertheless, it remains difficult to understand what Dublin managers mean by "resilience", as the measures taken do not differ from vulnerability reduction policies or prevention policies already tested in other territories. These measures in no way upset the conception of urban development, even if there are occasional reflections on the development of the districts of Ringsend and Spencer Dock. In reality, the planning logic remains the same, since, according to the managers, the consequences of flooding can be mitigated.

20Despite the emergence of a new vocabulary among managers and political actors, the mapping of actions carried out by the DCC (Fig. 4) illustrates that management approaches are not innovative. All the actions undertaken are concentrated in the areas likely to be flooded, even though the management plans are adapted to the scale of the catchment area as a whole. Everything happens as if development decisions taken outside floodplain areas had no consequences on floods and that only floodplain areas suffered the consequences of floods (whether they be economic, social impacts...). It therefore seems that the use of the concept of resilience by Dublin managers and political actors is not appropriate. Indeed, the use of this concept suggests that the policy in this area is innovative. In practice, however, the majority of actions undertaken are not.

Resilience to floods: different academic positions

21Dublin City is no exception. Other cities also have policies to improve the resilience of their territory in order to cope with floods (Mainz, Nijmegen, Stuttgart...) (Ashley et al., 2012). Resilience is "in" for both managers and scientists. However, in the academic field, the omnipresence of resilience questions its relevance (Lhomme, 2012).

22The different acceptances of resilience in the academic domain, linked in part to its polysemy, are illustrated by the different possible interpretations of the link between vulnerability and resilience. Indeed, the relationship between these two concepts is plural (Cutter et al., 2008). Some researchers then undertook to summarize in a didactic way various acceptances of this relationship (Cutter et al., 2008). One of the common acceptances opposes these two concepts. Within this framework, resilience and vulnerability would be two opposing sides of the same coin (Folke et al., 2002). Resilience would be a positive attribute of a system that would need to be increased, while vulnerability would be a negative attribute that would need to be decreased (Rufat, 2012). However, not all resilience is good to take (Reghezza et al., 2012). Resilience is therefore not necessarily a desirable property of systems. Moreover, this interpretation of the link between resilience and vulnerability leads to tautological reasoning: what does the concept of resilience bring, if it is only the positive decal of vulnerability (Rufat, 2012)?

23We therefore find in the academic field the problem raised by the analysis of the measures undertaken by Dublin managers. The opposition of these two concepts has been discussed and contested (Klein et al., 2003; Cutter et al., 2008; Dauphiné and Provitolo, 2007; Provitolo, 2009; Reghezza et al., 2012), but has been retained by a reference organization such as the Resilience Alliance.

24Depending on the sources, resilience will more or less encompass the different phases of risk management (Lhomme, 2012). Thus, from a concept limited to post-disaster (return to "normal"), associated with the idea of "rebound", restoration, recovery (Vale & Campanella, 2005), resilience can be considered as a concept encompassing all measures undertaken to reduce risks, before, during and after the crisis (Bruneau et al., 2003).

25In flood management, a first stream has a very broad vision of resilience. Therefore, "resilience measures" apply at different scales - from building to catchment - with diverse actions. The resilience of communities expected to cope with hazards is advocated, and the resilience of critical infrastructure is seen as imperative. To support this positioning, it is possible to rely on the actions undertaken by the various managers claiming to put the concept of resilience into practice on their territory. For some, resilience would thus be the modus operandi for implementing more sustainable risk management policies (Pasche & Geisler, 2005; Toubin et al., 2012). This vision can be found in European INTERREG projects involving academics and managers (FloodResilienCity and FREEMAN project). Researchers associated with these projects offer to study and assess resilience taking into account all parameters contributing to improved risk management (Serre, 2011; Schelfaut et al., 2011). This work, which is intended to be operational, is fed by the discourse and practices of managers. Some managers then claim to rely on this academic work to implement their resilience policy.

26In this framework, resilience is sometimes presented as a risk management policy, consisting no longer in fighting against hazards, but adapting to them in order to reduce their negative impacts to a minimum (Pasche & Geisler, 2005). A second stream gives a more restrictive interpretation of resilience. Resilience is therefore defined in reference to another strategy, that of resistance (De Bruijn, 2005). A system's response to flooding then depends on its resilience and resistance (De Bruijn, 2005; Serre 2011). This positioning recalls the work of Burton and White (1978). Indeed, the response of a system can be considered in two ways. First, a lack of response, as the flood was wiped out by the protection systems as part of a resistance strategy (a protection strategy). Secondly, the impacts caused by flooding require a response (evacuation, reconstruction, etc.), as part of a resilience strategy (De Bruijn, 2005).

27A resistance strategy then consists in not being flooded and a resilience strategy consists in recovering from a flood (De Bruijn, 2005). The strategies of resistance consist in fighting against floods in order to eliminate them, this strategy of full "protection" implies the implementation of structural measures. Resilience strategies recognize flooding and require preparation to cope, which involves non-structural measures. It is more about "living with floods" (Vis, 2003). This approach is closer to those who consider resilience in a logic of assimilation and not of opposition to hazards (Dauphiné & Provitolo, 2007). This second current is no longer centred on a logic of zero damage. Rather, it is centred on a logic of rapid recovery (which is not a return to the initial state since resilience requires a learning process to avoid repeating past errors) or continuity of operation despite damage.

28In summary, there are at least two academic trends in flood resilience. A first current considers resilience as encompassing all the approaches undertaken to manage risks, including traditional protection strategies (Serre, 2011). A second current tends to distinguish, or even oppose, protection strategies and resilience strategies, as risk management can combine these two approaches (Fig. 5). However, if Dublin managers refer to resilience in a fairly inclusive way, their management policy ignores recovery after a flood. This aspect is nevertheless present in the two academic currents presented and therefore seems to constitute resilience.

Analysis of Dublin City's network infrastructure

Positioning adopted to analyse the resilience of the city of Dublin to floods

29Two terms seem to emerge from the different definitions of resilience: absorption and recovery (Lhomme et al., 2010). The term absorption is widely used in ecology, while recovery is very present in risk studies to characterize the recovery of the systems studied.

30Based on this definition, an approach to understanding urban resilience to floods has recently been proposed (Lhomme et al., 2010; Serre, 2011). This general approach considers the city as a system. Indeed, resilience is a concept that applies to systems and can even be seen as a concept derived from the complexity paradigm leading to a preference for systems approaches. For this, it is possible to rely on an important literature defining the city as a system (De Rosnay, 1975 ; Pumain, 1997 ; Lajoie, 2007 ; Nabaa, 2011).

31Technical networks prove to be the support of relations between the different components of the urban system (companies, inhabitants, productions...). Ipso facto, the dysfunctions of these networks imply the isolation of certain components of the urban system and consequently a questioning of the functioning of the city. This suggests that these networks play an essential role in the mechanisms of risk diffusion between the different urban components (absorption capacity) and in the mechanisms of reactivation of the urban system (recovery capacity) (Lhomme et al., 2010; Lhomme, 2012). This hypothesis is corroborated by post-disaster experience feedback (Ponton et al., 1988 ; Rouzeau et al., 2010) and by analysis of technical network dysfunctions (Matisziw et al., 2008 ; Johansson & Hansel, 2010).

32While the analysis of the strengths and weaknesses of technical networks may resemble a vulnerability diagnosis, it does contribute to examining the resilience of the territories studied. There is in fact a continuum between these two concepts, with vulnerability centred on damage, while resilience focuses more on recovery, restoring systems (Fig. 6). However, as the damage to the technical networks conditions the recovery (the return to "normal", the return to service) of the territories, it partly determines their resilience. Similarly, within these networks, the damage suffered conditions their return to service and consequently their resilience. In this context, the two concepts are not opposed to each other and can hardly be separate d.

Network infrastructure in the face of past floods

33The city of Dublin is regularly exposed to rainfalls caused by an inability of the sewerage system to cope with rainfall (Leahy, 2009). The Docks area network is currently operating at full capacity (Corcoran, 2007). Two shortcomings were identified in this area: the undersizing of the pipelines and the insufficient capacity of the lift pumps. Any development in the Docks area can contribute to increasing urban runoff and therefore the frequency and magnitude of these already recurrent floods. A study produced in July 2002 concluded that the only viable solution involved the construction of a new lift station in this area (GDSDS, 2005).

34Since the 18th century, the sewerage network and the city of Dublin have developed jointly (GDSDS, 2005). This sanitation network is now centralised around a single treatment plant (Ringsend) which will soon reach its limits (GDSDS, 2005). This network will soon enter a new phase of development, which is necessary to cope with Dublin's development. The construction of a new treatment plant should thus curb this exacerbated centralization.

35Many streets can be flooded by stormwater (Leahy, 2009). Dublin's road network, which is already congested in "normal times", can become saturated during flood periods, as evidenced by the 2011 floods (Leahy, 2011). For example, travellers to Dublin airport were asked to check in advance whether their itineraries were practicable. In addition, despite many pumps being sent, part of the ring road was closed for more than 24 hours. Major disruptions were recorded on the M1 and M7 motorways (more than an hour late on a traffic already overloaded in normal times). In fact, the functioning of the Dublin road network is conditioned by that of the drainage network during flood periods.

36Dublin's power grid is pretty well meshed. Indeed, its port ensures an efficient supply of coal to the feeding stations located in the Docks district. In the south, a hydroelectric station was built at Pollaphuca in the Wiclow Mountains. Finally, processing stations such as Inchicore and Maynooth (in the East) or Finglas (in the North) are essential to ensure the connection of the Dublin city network to the entire national network (Combes & O'Shea, 2012). This network does not appear vulnerable to flooding. Indeed, the electricity network managers seem to be well aware of these hazards and there was no blackout during previous floods. At the national level, such an event is even considered unlikely. Nevertheless, nearly 1,200 people were without electricity on 27 October 2011 following preventative blackouts. In addition, these cuts may have affected infrastructure essential to the daily lives of Dubliners. For example, the Merrion Shopping Centre was without electricity for several days for security reasons (Leahy, 2011).

Dublin's technical networks in the face of a 200-year coastal flood

37There are several ways to approach technical networks. The choice was made to analyse the physical infrastructure, its functioning and its recovery in the face of a given hazard, taking into account the interdependencies existing between these networks. The work carried out is based on configuration analyses and not on flow analyses. In this section, analyses of the resistance capacities of sewerage and electricity networks are presented.

38This analysis is based on an original hazard: a two hundred year coastal flood (probability of 1/200 to occur each year). Indeed, given the likely effects of climate change, coastal flooding is attracting some attention from Dublin managers. In addition, this hazard affects part of Dublin's historic city centre.

39CCD managers provided us with the geographic data they had on the two networks under study. However, these data did not contain information on the vulnerability of these networks to flooding. Therefore, it proved necessary to make hypotheses in order to simulate the consequences of a flood. These assumptions give rise to two scenarios. The first scenario will be considered as the "main" scenario, as it is based on feedback from past floods in Dublin, where the sewerage network is considered very vulnerable and the electricity network as not very vulnerable. The second scenario will be considered as the "alternative" scenario, because it is based on a critical analysis of the first scenario and on a general knowledge of the vulnerability of technical networks to floods, the sewerage network is then considered as punctually vulnerable and the electricity network as globally vulnerable.

40In the main scenario, the sewerage system suffers significant damage (Fig. 7), while the electrical network does not suffer any damage. These results are a direct consequence of the assumptions made and can therefore be considered an integral part of the initial scenario. In a second stage, it is necessary to analyse how the networks react to this potential damage, while taking into account the possible interdependencies between the networks. This analysis highlights the opposition between, on the one hand, a disturbed sewer system and, on the other hand, an almost normally functioning electricity system. Indeed, the sanitation network, which is structured in a tree structure, is not able to cope with the identified damage. Thus, as damaged components are located near the Ringsend treatment plant, the malfunctions induced by these damages disrupt a large part of the sewer network (Fig. 7).

41First, the results of the alternative scenario differ from the results of the previous scenario. The electrical network appears to be segmented after the flood. The sewer system suffers much less damage compared to the previous scenario. The new assumptions seem to change the results. Nevertheless, in a second phase, given its network and the geographical positioning of the source substations, the electricity network does not suffer from major malfunctions. They are limited to parts that can be cut preventively. However, the situation is much more worrying for the sewerage network (Fig. 8). Indeed, the electrical network, very aggressive towards other technical networks, disrupts the sewerage network. For example, lifting pumps located near the Ringsend treatment plant are threatened by disturbances in the electricity grid. However, an important part of the sewerage network depends on the operation of these pumps. Finally, disturbances caused by the electricity grid on the sewerage network are problematic (Fig. 8).

42Therefore, various distributions of major issues can be analyzed in order to apprehend in part the resilience of the territory studied. The distributions of some major "recognized" issues (e.g. hospitals, fire stations) and some capital issues were analyzed (e.g. ministries, embassies) (Tab. 1). If the various issues analysed are concentrated outside the flooded areas, they will almost all be affected by the dysfunctions of the sanitation network (Tab. 1). On the other hand, the electricity network of the city of Dublin should be little disturbed taking into account its good meshing.

43To conclude, the Dublin flood management policy neglects the complex relationship between risk and territory, whereas an analysis of the resilience of this territory more precisely highlights this complexity.

Table 1: Summary of the results obtained concerning the distribution of certain issues regarding the functioning of the sanitation network.

	Fire stations	Hospitals	Departments	Embassies
Flooded areas	0 %	0 %	7,7 %	15,6 %
Disturbed areas scenario 1	75 %	90,6 %	100 %	100 %
Disturbed areas scenario 2	62,5 %	65,6 %	100 %	100 %

Serge Lhomme, 2012

Conclusion

44The difficulties in identifying the characteristics attached to the "resilience measures" promoted by the City of Dublin to cope with floods echo the different acceptances of resilience in the academic field. Dublin managers seem to view resilience as an antonym of vulnerability and a synonym of hazard control, as resilience combines both. This positioning refers to an encompassing vision of resilience, but neglects one of its main characteristics: the recovery (reactivation, restoration) of the territory following floods. In this context, Dublin's flood management policy is not innovative. Indeed, it is based on a classic logic of zero damage. It is therefore not based on an acceptance of the damage requiring a focus on the functioning of the territories during this damage (i.e. on the absorption capacities of these territories) and on the recovery of the territories (i.e. on the recovery capacities of these territories).

45In addition, the methodology developed highlights the lack of a systemic vision of Dublin's territory in terms of flood management. Indeed, the analysis of the resilience of the technical networks of the city of Dublin is not yet an integral part of the DCC's flood management policy, even though these networks are necessary for its proper functioning. Thus, in the case of a two hundred year flood, if almost all the major issues are located outside the flood zones, it emerges that these major issues will be very much affected by the malfunctions of the sanitation network. However, the DCC's policy does not take into account the complex link between risks and territories.

46The lack of a systemic vision of Dublin territory to cope with floods, even though resilience is a concept derived from the complexity paradigm, may seem surprising. This absence illustrates the difficulty for practitioners to take into account an unstable concept in the academic field. The vagueness surrounding resilience therefore questions its practical relevance in a tangible way. In any case, the analyses carried out on the technical networks of the city of Dublin are of interest to Dublin's managers and would not have been possible without their involvement, perhaps marking a turning point towards real measures aimed at improving the resilience of their territory.