Reg Environ Change (2014) 14:919–931

DOI 10.1007/s10113-013-0428-4

ORIGINAL ARTICLE

Climate adaptation of interconnected infrastructures:

a framework for supporting governance
L. A. Bollinger • C. W. J. Bogmans • E. J. L. Chappin • G. P. J. Dijkema •

J. N. Huibregtse • N. Maas • T. Schenk • M. Snelder • P. van Thienen •

S. de Wit • B. Wols • L. A. Tavasszy

Received: 29 June 2012 / Accepted: 16 February 2013 / Published online: 5 March 2013
 Springer-Verlag Berlin Heidelberg 2013

Abstract Infrastructures are critical for human society, comprehensively to facilitate adaptation governance. We
but vulnerable to climate change. The current body of suggest that it may help to ensure that the mental models of
research on infrastructure adaptation does not adequately stakeholders and the quantitative models of researchers
account for the interconnectedness of infrastructures, both incorporate the essential aspects of interacting climate and
internally and with one another. We take a step toward infrastructure systems. Further research is necessary to test
addressing this gap through the introduction of a frame- the framework in these contexts and to determine when and
work for infrastructure adaptation that conceptualizes where its application may be most beneficial.
infrastructures as complex socio-technical ‘‘systems of
systems’’ embedded in a changing natural environment. Keywords Climate change adaptation
Governance

We demonstrate the use of this framework by structuring Road
Electricity
Drinking water

potential climate change impacts and identifying adapta- Socio-technical systems
Systems of systems
tion options for a preliminary set of cases—road, electricity
and drinking water infrastructures. By helping to clarify the
relationships between impacts at different levels, we find Introduction
that the framework facilitates the identification of key
nodes in the web of possible impacts and helps in the Infrastructures are capital-intensive, long-lived, large-scale
identification of particularly nocuous weather conditions. systems that serve critical functions in support of human
We also explore how the framework may be applied more settlement and well-being. They are deeply embedded
within their environment and are constructed to operate
within particular ranges of environmental conditions. By
L. A. Bollinger (&)
E. J. L. Chappin

affecting the ‘‘normal’’ range of environmental conditions
G. P. J. Dijkema
M. Snelder
L. A. Tavasszy
Faculty of TPM, Delft University of Technology, and the frequency and severity of extremes, climate change
P.O. Box 5015, 2600 GA Delft, The Netherlands poses a potential threat to these systems—from degrading
e-mail: [email protected] their integrity and performance to inciting network-level
failure.
C. W. J. Bogmans
VU University Amsterdam, Amsterdam, The Netherlands The last decade has seen a shift in the research com-
munity from an exclusive focus on the role of infrastruc-
J. N. Huibregtse
N. Maas
M. Snelder
S. de Wit
tures in climate change mitigation toward recognition of
L. A. Tavasszy
potential vulnerabilities and the need for adaptation. This
TNO, Delft, The Netherlands
shift is reflected in numerous studies focusing on various
T. Schenk infrastructures, including water, electricity and transporta-
Massachusetts Institute of Technology, Cambridge, MA, USA tion (e.g., Decicco and Mark 1998; Hor et al. 2005; Kirshen
et al. 2008; Koetse and Rietveld 2009; Hunt and Watkiss
P. van Thienen
B. Wols
KWR Watercycle Research Institute, Nieuwegein, 2011; van Vliet et al. 2012). Studies such as these represent
The Netherlands an important step toward understanding the potential

123
920 L. A. Bollinger et al.

impacts of climate change on infrastructure and developing governance—in particular to facilitate the development of
suitable strategies for dealing with them. quantitative models and processes of stakeholder engage-
However, the current body of research is lacking in two ment. We conclude with a discussion and outlook.
key respects. First, it discounts the interconnectedness of
infrastructure components. The existing literature tends to
focus on the micro-level—the impacts on individual Framework
infrastructure components—and the macro/landscape
level—the effects on the natural systems surrounding The framework described in this section draws from two
infrastructures (Chappin and van der Lei 2012). A void is key theoretical traditions. First, we frame infrastructures as
left at the meso- or intermediate level—the level at which complex socio-technical systems—highly interconnected
the technical and social elements of infrastructures interact networks of interacting social and technical components
with one another and at which component impacts may that cannot easily be addressed independently from one
propagate into network-wide failures. another (Hughes 1987; Ottens et al. 2006; Simon 1973,
Second, much of the current scholarship disregards the 1962). Second, we frame infrastructures as systems of
interconnectedness of infrastructures with one another. systems—sets of heterogeneous, distributed systems
With a few exceptions (e.g., Kirshen et al. 2008; Hunt and embedded in networks at multiple levels that evolve over
Watkiss 2011), the existing literature tends to explore time (Agusdinata and DeLaurentis 2008). Important with
impacts and adaptation strategies associated with different respect to climate change is that these heterogeneous,
types of infrastructures separately. This approach disre- multilevel systems can furthermore be seen as embedded
gards possible interconnections between infrastructures, in within and heavily linked with their environment (Cash
particular the potential for disruptions within one infra- et al. 2003; Gunderson and Holling 2002).
structure system to spillover to others. The proposed framework for infrastructure climate
In this paper, we take a first step toward a more compre- adaptation is illustrated in Fig. 1. At the core of this
hensive approach—one that recognizes the interconnected- framework is a multilevel chain, beginning with climate
ness of infrastructures, both internally and with one another. change. Climate change translates into shifting extreme and
This first step takes the form of a framework for infrastruc- mean values for weather variables, as well as changes in (sea)
ture climate adaptation, which we introduce in the next water levels, hydrological cycles, soil conditions, vegetation
section. Following a description of this framework, we and other environmental conditions. It is a long-term phe-
demonstrate its use in structuring potential climate change nomenon playing out over a scale of decades or even cen-
impacts and identifying adaptation options for a preliminary turies, but its symptoms may be palpable on much shorter
set of cases—road, electricity and drinking water infra- timescales—weeks, days, hours or even minutes. These
structures. After this, we discuss how the framework may be symptoms are expressed in the form of loads/events such as
applied more comprehensively in support of adaptation droughts, heat waves, windstorms and floods (IPCC 2012).

Fig. 1 The proposed framework for infrastructure climate adaptation. network level within an infrastructure and the spillover of impacts
The framework captures several key aspects of climate–infrastructure between different infrastructures. The framework also highlights the
interactions—the relationships between climate scenarios and loads/ relevance of the environmental, social and technical domains and the
events and between loads/event and infrastructure components. It also interactions between them
captures the spread of impacts from the component level to the

123
A framework for supporting governance 921

Component impacts arise when these loads and events the set of occurrences surrounding a particular load/event.
encounter and affect the components of the socio-technical On Tuesday August 7, 2008, a severe rainstorm struck the
infrastructure—for example, a tunnel may flood, the cool- Botlek Tunnel, a key road tunnel in the Rotterdam harbor
ing system of a thermal power plant may run short on area of the Netherlands. The severity of this storm cannot
cooling water, and people may suffer from extreme heat. be directly tied to climate change, but it is representative of
Depending on various factors, these component-level the types of events that may occur with increasing fre-
impacts may affect the performance of the link or node quency as a consequence of climate change.
within which a component is embedded (e.g., the transport The storm directly affected a component of the elec-
corridor of which the tunnel is a part). Due to a breach in tricity infrastructure and quickly spread to a key compo-
integrity, this link or a node may cease to function prop- nent of the transport infrastructure. A lightning strike to
erly—traffic in a transport corridor may become congested, electrical circuitry near the Botlek Tunnel cut power to a
and a power plant may be forced to reduce output or shut set of pumps that are normally used to drain excess water
down entirely. from the tunnel. Due to the failing of these pumps and the
From the level of a link or node, a disruption can heavy rainfall—45 mm in a five-hour period—the water
propagate through the network. Depending on conditions, a level in the tunnel rose to a height of one meter in some
single breach may lead to congestion elsewhere, over- areas. These impacts at the component level spread
loading other links and eventually causing a network-level throughout the network, causing a number of further
failure—for example, traffic gridlock, a cascading failure impacts at various levels:
in the electricity grid. Such failures typically have sub-
• At the link level, the incident resulted in traffic jams and
stantial economic impacts (Laird et al. 2005). The magni-
subsequent vehicle loss hours on the A15 extending
tude of these impacts depends on the degree to which
15 km in both directions (Rosmuller et al. 2011).
additional network effects occur, the sensitivity of the
• At the network level, the traffic jams on the A15
processes associated with network usage and the avail-
resulted in spillback effects in the form of increased
ability of options to use other networks.
traffic on the A4 (Benelux Tunnel) and extra travel time
In this context, it is important to keep in mind that any
on other routes, including veer Rozenburg, Spijkeniss-
given infrastructure is not an isolated technical system—it
erbrug, N57, A29 (Heinenoord Tunnel) and A16
is part of a system of infrastructure systems. Components
(Drecht Tunnel).
of one infrastructure may be linked with components of
• In other networks, the incident resulted in extra travel
another. Road signals require electricity; power plants
time for travelers who shifted to other modes, as well as
need transport routes for fuel delivery; and rail systems
the diversion of inland ships from the Botlek Bridge.
depend on telecommunications. Disruptions at the com-
ponent, link/node or network level in any single infra- Translated into economic terms, the combination of
structure system may propagate to other infrastructure these phenomena resulted in costs for those delayed in
systems. traffic and for those forced to take other modes, as well as
This system of infrastructure systems is embedded for those who decided to depart later or cancel their trips
within broader social, economic and political systems. altogether. A rough estimate suggests that the economic
Through processes of governance, these systems engender cost of this incident was 367,500 euros (Rosmuller et al.
adaptation measures. These measures may be directed at 2011).
various points in the chain of an infrastructure system. The impacts of this incident would have been larger
Measures can be directed at the component level—the were it not for several measures already in place. Two
points of interaction between environmental variables and chief adaptation measures in this case were the deploy-
infrastructures. They can also address the link/node level— ment of mobile pumping units to the Botlek Tunnel and
preventing component impacts from affecting the perfor- the closing of the Botlek Bridge to inland shipping so as
mance of individual links or nodes (robustness). Or, they to allow for greater volume of automobile traffic. Given
can be aimed at the network level, accepting that individual the costs of the incident, the sufficiency of existing
links or nodes may fail but ensuring that the network is able measures to deal with such loads/events was questioned
to accommodate these failures and continue functioning and a governance process was initiated (Rijkswaterstaat
(resilience). 2009).
The governance process in this case was reactive—that
Example—flooding of the Botlek Tunnel is, initiated in reaction to the incident in question. In this
paper, we suggest that the proposed framework can support
To further clarify the proposed framework for infrastruc- the proactive identification of potential climate change
ture climate adaptation, we use this framework to structure impacts and adaptation options.

123
922 L. A. Bollinger et al.

Application of the framework—structuring impacts Structuring possible impacts

and identifying adaptation options
Using the proposed framework as a guide, Fig. 2 summa-
In this section, we use the framework introduced above to rizes the impact chain of several extreme weather condi-
facilitate the structuring of potential climate change tions. For example, heavy rain results in reduced vision and
impacts on three different infrastructures—road, electricity thus reduced speeds and reduced capacity. If demand is
and drinking water. In each case, we present this struc- high enough, this will result in congestion on all affected
turing as a multilevel causal web and use it to facilitate the links with high demand in the network. The situation may
identification of possible adaptation measures. worsen because different links influence each other, caus-
ing congestion to spill back to other links and other parts of
Road infrastructure the network. The potential for serious disturbances is
especially acute in densely populated regions where a
Road networks are vulnerable to various types of distur- single event can cascade easily through the network.
bances—extreme wind, flooding, extreme temperatures and
droughts. These types of events may affect the infrastructure Identifying adaptation options
in very different ways, some temporary and some extended
(Snelder 2010). Extreme wind, for instance, may spread Drawing from the impacts web visualized in Fig. 2, we can
debris on roads, slowing traffic and reducing road capacity in identify possible adaptation measures at different levels. At
the short term. Extreme temperatures, on the other hand, may the component or link/node level, important adaptation
cause rutting and melting of asphalt, resulting in longer-term measures could include optimizing or redesigning compo-
issues and necessitating maintenance. nents. In cases of extreme rain, options include increasing

Fig. 2 Anticipated climate change impacts on road infrastructures network impacts; and black boxes represent social and economic
(draws from research by Oostroom et al. 2008; Koetse and Rietveld impacts. Boxes represent impacts on technical infrastructure compo-
2009; SWOV 2009; TRB 2008). In line with the terminology of nents, and ovals represent impacts of climate events on infrastructure
Fig. 1, white boxes represent climate events; light gray boxes functions
represent component, node or link impacts; dark gray boxes represent

123
A framework for supporting governance 923

the capacity of drainage and storage (pump cellars) and long-term increases in peak electricity loads (Rothstein
increasing the pump capacity to pump the water from the et al. 2008). With respect to electricity transmission and
pump cellars out of the system. These measures could distribution, higher temperature extremes are expected to
either be taken during maintenance or be included in new increase resistance and sag in overhead lines, and droughts
designs. Melting and rutting of asphalt during periods of may reduce the capacity of underground cables (Rade-
extreme temperature could be addressed through the maekers et al. 2011).
development and implementations of alternative road sur-
faces. Measures such as these could help to alleviate node- Structuring possible impacts
level issues such as closed or blocked roads and bridges
and reduced speeds and road capacities. Using the proposed framework as a guide, Fig. 3 summa-
At the network level, adaptation options entail both rizes the impact chains of several extreme weather
measures to improve the design of the road network and conditions.
measures to improve the management of the road network
under extreme conditions. As suggested by the structuring Identifying adaptation options
in Fig. 2, key network design measures could include
creating more route alternatives (redundancy) so as to Drawing from Fig. 3 and from available literature, we
allow drivers to avoid closed, blocked or congested routes, identify several relevant adaptation measures at different
as well as introducing buffers and unbundling. Network levels. At the component or link/node level, an important
management measures could include weather alarms and category of adaptation measures for electricity infrastruc-
the development of incident management scenarios for tures could include modification of generator designs to
different extreme weather events. improve performance under extreme conditions such as
Measures directed at the infrastructure’s environment droughts and extreme wind speeds and temperatures. New
entail both the optimization of environmental conditions thermal power plants can incorporate closed circuit cooling
(e.g., modification of drainage patterns, subsoil composi- systems (Tzimas 2011), and old ones can be retrofitted with
tions and vegetation) and location choice in network cooling towers. Furthermore, renewable installations such
design. These types of measures could also be incorporated as hydropower dams and wind parks can be designed to
into spatial planning guidelines. Measures in this category take into account uncertainty concerning future climate
could also incorporate strategies for enhancing coopera- conditions, such as more variable precipitation and altered
tion/communication between traffic managers and other wind patterns/speeds. In addition to design and retrofitting
authorities, such as emergency services. measures, adaptation may be directed at the management
Drawing from Fig. 2, we can also see that no single of individual nodes. For instance, power plant maintenance
category of load or event causes a disproportionate set of operations can be scheduled in the summer, or cooling
impacts—various types of loads/events are relevant. This water regulations can be relaxed during crises to avoid
suggests that adaptation strategies cannot be productively capacity shortages. Measures such as these can help to
geared toward dealing with a particular type of event, but mitigate generation shortages, which can lead to blackouts
should be structured to deal with diverse types of events. and other disruptions at the network level.
At the network level, three distinct types of measures
Electricity infrastructure can be identified. First, adaptability measures can be used
to improve the capacity of electricity infrastructures to
The supply, demand, transmission and distribution of actively respond to environmental changes. This category
electricity will be affected in myriad ways by a changing includes demand-side management—improving the
climate. With respect to electricity generation, increases in responsiveness of loads to the availability of generation
mean air and water temperatures and decreases in river capacity—as well as dynamic rating (Tennet 2010), self-
flows are likely to affect the availability and efficiency of healing grid mechanisms and islanding techniques (Mili
thermal generators, as well as the outputs of hydropower 2011). Second, diversity measures can improve the diver-
and other renewable energy generation technologies (Koch sity of generation technologies in an infrastructure net-
and Vogele 2009; Linnerud et al. 2011; Mideksa and work. As illustrated in Fig. 3, different generation
Kallbekken 2010). With respect to electricity demand, technologies have different vulnerabilities to climatic
climate change is anticipated to result in reduced demand variables, so a technologically diverse generation portfolio
for electric heating and may increase demand for air-con- can improve the likelihood that infrastructure is able to
ditioning and refrigeration (Petrick et al. 2010). Further- meet demand under abnormal circumstances. Third,
more, there is evidence that extreme weather events may redundancy measures can ensure sufficient slack and
induce the purchase of cooling devices and subsequent backup capacity, both in electricity generation and in the

123
924 L. A. Bollinger et al.

Fig. 3 Anticipated climate change impacts on electricity infrastruc- Pryor et al. 2005; Pryor and Barthelmie 2010; De Groot et al. 2006;
tures (draws from research by Rothstein and Parey 2011; Rademae- Rothstein et al. 2008)
kers et al. 2011; Wilbanks et al. 2008; Mideksa and Kallbekken 2010;

electricity grid. Generation capacity mechanisms can electricity infrastructure under cases of extreme tempera-
include a variety of possible measures to incentivize tures may be especially important from a governance
capacity investments—capacity subscriptions, reliability perspective.
contracts, capacity payments and strategic reserves (Finon
and Pignon 2008). Drinking water infrastructure
Similar to the road infrastructure case, measures direc-
ted at the infrastructure’s environment can include efforts Climate change may impact the functionality of drinking
to modify the environment or efforts relating to the water infrastructure indirectly by altering soil properties
placement of infrastructure components within the envi- and soil movement. The deterioration of pipe systems
ronment. Efforts to modify the environment might include results from failures in the materials and/or deterioration of
measures to ensure the regular trimming of vegetation in the construction. Various types of climate change-related
the vicinity of overhead power lines or to enhance the flood extreme weather conditions can affect these deterioration
defenses around substations or power plants. Measures processes. Droughts and high temperatures can dry out and
concerning component placement might include legislation shrink the soil, causing an increase in pipe breakage rates
or guidelines to incentivize the construction of thermal (Newport 1981; Kleiner and Rajani 2001). Furthermore,
power plants in locations with ample cooling water supply high temperatures often result in higher water demand
(e.g., coastal locations). (Billings and Jones 2008), which can increase water pres-
Like in the road infrastructure case, we can see that a sure close to pumping stations and raise flow velocities. In
diversity of loads/events may have impacts on the elec- turn, these effects can increase the risk of water hammer
tricity infrastructure. However, in this case, we can also see and associated damage to pipes. Higher soil temperatures
the particular threat posed by extreme temperatures. This may also affect drinking water temperature, which can
suggests that adaptation measures to shore up the adversely affect quality.

123
A framework for supporting governance 925

Structuring possible impacts be altered to increase the structural flexibility of the net-
work to withstand differential settlements.
Figure 4 gives an overview of how climate events can At the network level, adaptation measures can address
impact drinking water distribution systems. All climate both the design of the network and management of the
events may change the loads or strength of the pipe, which network. Even though currently most networks are looped
may result in pipe (or joint) failure. A small part of the on all levels, a branched network design for the tertiary
network will then be cut off from the water supply for network (neighborhood level, e.g., Vreeburg et al. 2009)
repairs. The resulting economic impacts include the costs offers benefits both for water quality (especially but not
associated with repairing both the distribution network and exclusively when soil temperatures increase) and, contrary
other impacted infrastructure in the surroundings of the to common belief, for continuity of supply in case of pipe
burst, and interrupted water supply. bursts (Vreeburg et al. 2009). The number of connections
affected by a pipe burst is determined by the size of sec-
Identifying adaptation options tions which can be isolated, or in other words the valve
density in a network (and of course also their locations).
At the component or link/node level, important adaptation Either the valve density or the size of tertiary network
measures for drinking water infrastructures include the branches can be adjusted to cope with possibly increasing
materials used for the construction of pipes and the types of failure rates.
joints applied between them. The centrality of increased Similar to the electricity infrastructure, demand-side
pipe loadings in Fig. 4 suggests that a key adaptation may management strategies can also be beneficial and can be
be the implementation of pipe materials with increased executed using various technical, legal and economic
mechanical strength (e.g., steel pipes) or an increased mechanisms (Niemczynowicz 1999). However, in contrast
flexibility to withstand differential settlements induced by to the electricity infrastructure, demand-side measures
climate change (e.g., PVC pipes). In areas expected to cannot be based on the price elasticity of the water (at least
experience groundwater salinization due to sea-level rise, in developed countries), since costs of drinking water are
noncorroding materials have to be selected. Secondly, the low compared to those of energy. Effective short term
types of joints between the pipes or the joint distance can demand-side measures are hosepipe bans as used in the UK

Fig. 4 Anticipated climate change impacts on drinking water infrastructures (draws from research by Hu and Hubble 2007; Newport 1981;
Kleiner and Rajani 2001; Rajani and Tesfamariam 2004; Billings and Jones 2008; Rajani and Kleiner 2001; Van Daal et al. 2008)

123
926 L. A. Bollinger et al.

and Australia during severe droughts. Improved monitoring stakeholders and quantitative models. Adaptation gover-
of the infrastructure, for example, by estimating stresses on nance may be seen as a process of stakeholders interacting
buried infrastructure due to soil movements using satellite with the best possible data while managing inherent uncer-
observations (Dheenathayalan et al. 2011), can help to tainties to determine specific measures for implementation.
quickly locate and repair infrastructure damage. Models serve to inform these stakeholders about the
Measures directed at the infrastructure’s environment behavior of the systems in question and possible conse-
entail, most importantly, measures to mitigate risks to pipe quences of various courses of action. While our focus here is
integrity. For instance, the soil around a pipe can be on quantitative models, it is important also to point out the
modified to reduce soil differential settlements. Also, relevance of softer techniques (e.g., models using possible
avoiding trees located near the pipes may reduce the future scenarios without probabilities attached), which may
chance of pipe bursts when a tree is uprooted by the wind. also be combined with rigorous quantitative techniques. In
either case, however, these models must be grounded in the
Synthesis particular geographies and socio-political realities within
which the infrastructure in question is situated.
In the previous sections, we have applied the proposed In this section, we explore the use of the proposed
framework to facilitate the structuring of possible climate framework in adaptation governance in practice. In doing
change impacts and the identification of adaptation impacts so, we take into account both the need for framing and
for three different infrastructures. The causal web diagrams analysis to ground the risks and possible responses, and the
illustrate that the impacts of climate change on infra- potential consequences of interconnections between dif-
structures should not be seen in static terms—impacts at ferent infrastructures. Insofar as it provides a coherent
the level of components may radiate to higher levels, conceptualization of interacting climate and infrastructure
dynamically interact with other impacts and spread to other systems, the proposed framework can facilitate both
networks. These diagrams also help us to see several key stakeholder engagement processes and the development of
impacts—impacts with a diversity of causes—as well as quantitative models. In particular, the framework high-
particularly nocuous weather conditions—conditions with lights the essential aspects of interacting infrastructure and
a diversity of impacts. In the case of both electricity and climate systems that should be considered in adaptation
drinking water infrastructures, extreme temperatures governance processes. These aspects are listed in Table 1.
appear to be an important threat. In the case of road
infrastructure, the threats are more dispersed. Table 1 Key relationships highlighted by the proposed framework
The causal webs point to several key adaptation strate- that should be captured in adaptation governance of infrastructures
gies and illuminate strategies that may be applicable across Relationship Description
different types of infrastructures. For instance, the
Relationships between climate A range of climate change
centrality of increased pipe loadings in Fig. 4 points to
scenarios and loads/events scenarios and their consequences
increased pipe strength and flexibility as a potential key on the assumed frequencies of
adaptation. Furthermore, the diversity of impacts caused by various types of loads/events
extreme temperatures in the case of both the electricity and Relationships between weather The (potentially numerous)
drinking water infrastructures suggests that a diversity of variables and infrastructure relationships between weather
measures may be appropriate under such circumstances. In components variables and infrastructure
components—for example, the
the case of both infrastructures, this may include demand- relationship between
side management combined with improved system moni- precipitation and driver visibility/
toring. In addition to facilitating the identification of speed, or the relationship
adaptation options, the proposed framework has provided between temperature and
electrical resistance in power
us with a language for conceptualizing different types of lines
infrastructures such that they become comparable and Relationships between the Key interactions, both social and
compatible with one another. This is a first step en route to components, links and nodes technical, that may cause
governance processes that effectively address the inter- of an infrastructure disruptions to spread from the
connectedness of infrastructures. component level to the link/load
and network levels
Relationships between different The interconnections between
infrastructures different infrastructure networks
Supporting adaptation governance (e.g., road, rail and electricity)
that may cause disruptions to
Our application of the framework has thus far disregarded spillover from one network to
another
several key aspects of adaptation governance—in particular

123
A framework for supporting governance 927

Supporting stakeholder engagement Systems models may be constructed solely by experts

based on empirical research, collaboratively with stake-
A critical element of effective governance is involving a holders, using computer-driven agent-based modeling, or
range of stakeholders with varying interests and perspec- via some combination of these approaches (Costanza and
tives (Functowicz and Ravetz 1993; National Research Ruth 1998; Janssen and Ostrom 2006). Collaborative
Council 2009). In a world of post-normal science, decision- approaches allow for a broader range of insights and
makers and scientists become coproducers of knowledge perspectives. Users must appreciate that their models are
(Functowicz and Ravetz 1993; Webb 2011). Stakeholder inherently simplifications and can overlook key elements
engagement done poorly can, however, lead to frustration, because their importance is not appreciated, is misunder-
wasted time and resources, and ultimately project failure. stood or miscalculated or is not apparent at the time of
Fortunately, there exists a wealth of knowledge on how model construction. This may be particularly problematic
stakeholder engagement can be done well. Best practices when considering climate change, given the significant
include using professional neutral facilitators; engaging the uncertainty.
full breadth of stakeholders from both inside and outside of Another relevant technique is collaborative adaptive
government; engaging stakeholders on an ongoing basis management. Adaptive management presupposes that,
rather than only late in the game when key decisions are because of uncertainty and changing conditions like those
being made; and researching and exploring data together associated with climate change, no plan or project will be
via joint fact-finding procedures (Innes and Booher 2010; optimal from the outset and remains so throughout its life
Susskind and Crump 2009). cycle. Instead, ongoing monitoring and evaluation is con-
Stakeholder engagement is most effective when it is ducted, and both designs and management systems are left
deeply entrenched in the institutional fiber of decision- flexible so that new information can be used to iteratively
making (Margerum 2011). This requires support and hone practice (Doremus et al. 2011; Gunderson and Holling
resources from higher levels of government, including 2002; Susskind and Crump 2009; Williams et al. 2009).
legislative changes (Camacho 2009). Knowledge and Experts may conduct adaptive management technocrat-
decision-making around infrastructure management can no ically, but the presence of irreconcilable uncertainties and
longer be treated as the reified domain of experts, but rather value-based decisions make it more effective when man-
must be brokered between experts, policy makers and other aged by multistakeholder groups—that is, as collaborative
stakeholders. adaptive management (Williams et al. 2009; Innes and
We argue that the proposed framework can facilitate Booher 2010). Adaptive management can appear expensive
processes of stakeholder engagement and interaction in up front compared to traditional decision-making, but often
adaptation governance. Each stakeholder enters such pro- results in much more efficient, cost-effective and success-
cesses with a unique set of mental models—that is, mental ful management over the long term (Doremus et al. 2011).
representations of various situations and systems—and Given the dynamic and uncertain nature of climate change,
interests that form the basis for his or her argumentation this may be all the more true. Adaptive management pro-
and reasoning in decision-making (Susskind and Crump vides a way forward while acknowledging that we are most
2009). Application of the proposed framework can help to likely to get our forecasts wrong and thus must continue to
ensure that the mental models and interests of stakeholders monitor, evaluate and adapt as we proceed.
incorporate the essential aspects of the system in question
(Table 1) and subsequently that these aspects are incor- Supporting the development of quantitative models
porated into decisions that lead to the development of
adaptation measures. Infrastructures are complex socio-technical systems of sys-
Several techniques can be used to support facilitated tems. As demonstrated above, the impacts of extreme
decision-making processes in complex environments. One weather events may cascade through different levels of an
of these is systems thinking. Systems thinking promotes a infrastructure network, affecting the myriad decisions made
mind-set in which stakeholders challenge their mental by various actors and spilling over to other infrastructures.
models and explore the complex web of interactions in the Together with the massive uncertainties associated with
relevant system(s) so that they may identify the most climate change, the complex sets of interactions triggered by
effective intervention points and anticipate the likely range such events challenge our cognitive capabilities and com-
of consequences (Meadows 1999). Models or diagrams are plicate the selection of effective adaptation measures.
typically constructed to map the key elements of the system Modeling and simulation techniques can help stakeholders to
and how they relate (Sterman 2000). Positive and negative navigate this complexity, playing a key role in facilitating the
feedback loops are given particular attention, as they drive identification of effective adaptation measures (Claussen
stability and change in systems. et al. 2001; Larsen et al. 2007; Stern 2007).

123
928 L. A. Bollinger et al.

The proposed framework highlights the essential aspects system boundaries and/or underlying assumptions and
of interacting infrastructure and climate systems that asked to discern an acceptable course of action in discus-
should be considered in developing models to support sion with one another. While such approaches inevitably
adaptation governance. Many models capturing the indi- introduce added subjectivity into the governance process
vidual relationships listed in Table 1 already exist. Models (although real objectivity is never possible in such situa-
of traffic and road congestion are widespread, and power tions), they may also help to more fully incorporate the
flow models are commonly used to anticipate congestion in complexity of the system in question into governance
electricity grids (e.g., Bando et al. 1995; Lobato et al. processes, in particular the consequences of interconnec-
2004). Numerous global climate models (GCMs) exist, and tions between infrastructures.
some have been, or are in the process of being, downscaled
to provide regional-level results (e.g., Frei et al. 2006; Synthesis
Giorgi et al. 1993). The vulnerability of road networks has
been evaluated using risk assessments through which the By highlighting the essential aspects of interacting infra-
probability of unwanted events and related consequences structure and climate systems, we suggest that the proposed
are assessed (Baarse et al. 2008; Bles et al. 2010). Con- framework can facilitate both stakeholder engagement
siderable work has also been done in elaborating the processes and the development of quantitative models in
component-level impacts of weather events (e.g., Rade- support of adaptation governance. Quantitative models are
maekers et al. 2011; Koetse and Rietveld 2009). an important ingredient in helping stakeholders to under-
While models such as these can provide valuable stand interacting infrastructure and climate systems, and
insights to support adaptation governance, they are insuf- much progress has been made in the development of
ficient for isolation. Some progress has been made in the models to inform adaptation processes.
development of integrative models—computational models However, quantitative models alone are not sufficient.
that, via direct or indirect linkage, allow for studying Interacting climate and infrastructure systems are charac-
multiple systems as an integrated whole. An example here terized by persistent uncertainty, fragmented knowledge
is a model developed by Van Vliet et al. (2012), which and locus of control, and the inherently subjective nature of
combines a hydrological model with an electricity pro- many decisions. Adaptation is thus most effective and
duction model and GCM outputs to arrive at conclusions efficient when multiple stakeholders are engaged such that
concerning the vulnerability of regional electricity supplies they can advocate for their interest and explore the impli-
to climate change. Such models are increasingly feasible cations of the various decisions across networks. Used in
from a computational standpoint, but are difficult to concert with the proposed framework, a variety of tech-
develop given the degree of coordination necessary to niques can make facilitated multistakeholder planning
accurately capture the interactions, even in the form of processes more effective.
static outputs. Moreover, such models still offer only a
partial view of the system in question. The model of Van
Vliet et al. (2012), for instance, does not capture the spread Conclusions
of node-level impacts to the network level (meso-level
interactions) and ignores interactions between different The adaptation of infrastructures is a key challenge posed
types of infrastructures. Models incorporating multiple, by our changing climate. In addressing this challenge, it is
interacting infrastructure systems exist, but are in their essential that governance processes adequately account for
nascence (Haimes and Jiang 2001; Panzieri et al. 2004; the interconnectedness of infrastructures, both internally
Pederson et al. 2006; Carreras et al. 2007; Rosato et al. and with one another. With the aim of supporting this, we
2008). Such models are nonexistent in adaptation literature. have introduced a framework for infrastructure climate
The development of integrative computation models is adaptation, which captures the key relationships of inter-
not the only approach to addressing inherent complexity in acting climate and infrastructure systems.
the interactions between infrastructure and climate sys- We have demonstrated how this framework can be
tems. An alternative approach involves the participation of applied to different types of infrastructures, and how it can
stakeholders in the integration of model results. For cases facilitate the structuring of possible impacts and the iden-
in which two models are not directly compatible due to tification of adaptation options. By helping to clarify the
different assumptions, system boundaries, etc., stakehold- relationships between impacts at different levels, the
ers may use their collective knowledge and mediated per- framework can facilitate the identification of key nodes in
spectives to facilitate the translation of the results of one the web of possible impacts—for example, increased pipe
model for use in another. Alternatively, stakeholders may loadings in the drinking water infrastructure. It can also
be exposed to the results of multiple models with different help to identify particularly nocuous weather conditions,

123
A framework for supporting governance 929

for instance extreme temperatures in the case of the elec- Bando M, Hasebe K, Nakayama A, Shibata A, Sugiyama Y (1995)
tricity infrastructure and the drinking water infrastructure. Dynamical model of traffic congestion and numerical simulation.
Phys Rev E 51(2):1035–1042
Insights such as these can support the identification of key Billings RB, Jones CV (2008) Forecasting urban water demand.
adaptation measures—for example, construction of drink- American water works association, USA
ing water pipes with stronger and/or more flexible mate- Bles T, Yves E, Fadeuihe JJ, Falemo S, Lind B, Mens M, Ray M,
rials—and adaptation measures that may apply across Sandersen F (2010) Risk management for roads in a changing
climate a guidebook for the RIMAROCC method final version
multiple infrastructures. ERA-NET ROAD. Available online at: http://wwwswedgeose/
Our application of the framework in the first part of this upload/FoU/Rimarocc%20Handbook%20complete%2020100930%
paper leaves out several key aspects of adaptation governance 20final%20version_corrected%2010pdf
in practice and largely disregards the role of interconnections Camacho AE (2009) Adapting Governance to Climate Change:
Learning to Manage Uncertainty. Notre Dame Law School Legal
between infrastructures. In the second part of the paper, we Studies Research Paper No 09-06
have explored how the framework may be applied more Carreras BA, Newman DE, Gradney P, Lynch VE, Dobson I (2007)
comprehensively to facilitate the development of quantitative Interdependent Risk in Interacting Infrastructure Systems. Pro-
models and processes of stakeholder engagement in a context ceedings of the 40th Hawaii International Conference on System
Sciences 0:112c
of interacting infrastructures. We argue that the proposed Cash DW, Clark WC, Alcock F, Dickson NM, Eckley N, Guston
framework can help to ensure that the mental models and DH, Jager J, Mitchell RB (2003) Knowledge systems for
interests of stakeholders incorporate the essential aspects of sustainable development. Proc Natl Acad Sci USA 100(14):
interacting climate and infrastructure systems. Techniques 8086–8091
Chappin EJL, van der Lei T (2012) Modeling the adaptation of
such as systems thinking and collaborative adaptive man- infrastructures to prevent the effects of climate change—an
agement can facilitate this. With respect to quantitative overview of existing literature. In: Third international engineer-
models, we similarly suggest that the proposed framework ing systems symposium—design and governance in engineering
can help to ensure the inclusion of the essential aspects of systems—roots trunk blossoms
Claussen E, Cochran VA, Davis DP (2001) Climate change: science,
interacting infrastructure and climate systems. Some progress strategies and solutions. Brill Academic Publishers, Arlington
has been made in the development of integrative models Costanza R, Ruth M (1998) using dynamic modeling to scope
capturing several of these aspects, but more work is neces- environmental problems and build consensus. Environ Manag
sary—in particular in the development of models capturing 22(2):183–195
De Groot RS, van Ierland EC, Kuikman PJ, Nillesen EEM, Platteeuw
interactions between different infrastructures. M, Tassone VC, Verhagen AJA, Verzandvoort-van Dijck S
In the preceding sections, we have pointed to a path for (2006) Climate adaptation in the Netherlands. Wageningen UR,
addressing the interconnectedness of infrastructures in The Netherlands
climate change adaptation. Much is left to do. We have Decicco J, Mark J (1998) Meeting the energy and climate challenge
for transportation in the United States. Energy Policy
suggested how the proposed framework can be useful in 26(5):395–412
the context of facilitating governance processes, but it is Dheenathayalan P, Cuenca MC, Hanssen R (2011) Different
not clear when and where its application may be most approaches for PSI target characterization for monitoring urban
beneficial. We have mentioned the limitations of quanti- infrastructure, In: 8th International Workshop on Advances in
the Science and Applications of SAR Interferometry. Frascati
tative modeling in capturing all the aspects of the proposed (Italy)
framework. But we do not know exactly how these models Doremus H, Andreen WL, Camacho A, Farber DA, Glicksman RL,
can best facilitate governance processes that must consider Goble D, Karkkainen BC, Rohlf D, Tarlock AD, Zellmer SB,
all of these aspects in combination. A key tenet of com- Jones S, Huang Y (2011) Making Good Use of Adaptive
Management Center for Progressive Reform. White Paper #1104
plexity theory is that we cannot fully know the path before Finon D, Pignon V (2008) Electricity and long-term capacity
us until we have walked it. The best we can do is to make adequacy: the quest for regulatory mechanism compatible with
sure we have the tools we might need along the way. electricity market. Util Policy 16(3):143–158
Frei C, Schoell R, Fukutome S, Schmidli J, Vidale PL (2006) Future
Acknowledgments This work is supported by the Knowledge for change of precipitation extremes in Europe: intercomparison of
Climate program, project INCAH—Infrastructure Networks Climate scenarios from regional climate models. J Geophys Res
Adaptation and Hotspots. 111:D06105
Functowicz SO, Ravetz JR (1993) Science for the post-normal age.
Futures 25(7):739–755
Giorgi F, Marinucci MR, Bates GT (1993) Development of a second-
References generation regional climate model (REGCM2) Part I: boundary-
layer and radiative transfer processes. Mon Weather Rev
Agusdinata and DeLaurentis (2008) Specification of system-of- 121(10):2794–2813
systems for policymaking in the energy sector. Integr Assess J Gunderson LH, Holling CS (eds) (2002) Panarchy: understanding
8(2):1–24 transformations in human and natural systems. Island Press,
Baarse G, Noordam D, Zanting HA (2008) De risicobenadering voor Washington, DC
adaptatie van infrastructuur aan klimaatverandering Raad voor Haimes YY, Jiang P (2001) Leontief-based model of risk in complex
Verkeer en Waterstaat. Arcadis interconnected infrastructures. J Infrastruct Syst 7:1–12

123
930 L. A. Bollinger et al.

Hor CL, Watson SJ, Majithia S (2005) Analyzing the impact of Ottens M, Franssen M, Kroes P, Van De Poel I (2006) Modelling
weather variables on monthly electricity demand. IEEE Trans infrastructures as socio-technical systems. Int J Crit Infrastruct
Power Syst 20(4):2078–2085 2(2–3):133–145
Hu Y, Hubble DW (2007) Factors contributing to the failure of Panzieri S, Setola R, Ulivi G (2004) An agent based simulator for
asbestos cement water mains. Can J Civ Eng 34(5):608–621 critical interdependent infrastructures. Securing Critical Infra-
Hughes TP (1987) The evolution of large technological systems. In: structures, Grenoble
Bijker WE, Hughes TP, Pinch TJ (eds) The social construction of Pederson P, Dudenhoeffer D, Hartley S, Permann M (2006) Critical
technological systems: new directions in the sociology and Infrastructure Interdependency Modeling: a survey of U.S. and
history of technology. MIT Press, Cambridge, pp 51–82 International Research. Idaho National Laboratory, US
Hunt A, Watkiss P (2011) Climate change impacts and adaptation in Petrick S, Rehdanz K, Tol RJ (2010) The Impact of Temperature
cities: a review of the literature. Clim Chang 104(1):13–49 Changes on Residential Energy Consumption. Kiel Working
Innes JE, Booher DE (2010) Planning with complexity: an introduc- Papers 1618, Kiel Institute for the World Economy
tion to collaborative rationality for public policy new. Routledge, Pryor SC, Barthelmie RJ (2010) Climate change impacts on wind
New York energy: a review. Renew Sustain Energy Rev 14(1):430–437
IPCC (2012) Managing the risks of extreme events and disasters to Pryor SC, Barthelmie RJ, Kjellstroem E (2005) Potential climate
advance climate change adaptation. In: Field CB, Barros V, change impact on wind energy resources in northern Europe:
Stocker TF, Dahe Q, Dokken DJ, Plattner GK, Ebi KL, Allen analyses using a regional climate model. Clim Dyn
SK, Mastrandrea MD, Tignor M, Mach KJ, Midgley PM (eds) 25(7–8):815–835
Cambridge University Press, Cambridge Rademaekers K, van der Laan J, Boeve S, Lise W (2011) Investment
Janssen MA, Ostrom E (2006) Empirically based agent-based models. needs for future adaptation measures in EU nuclear power plants
Ecol Soc 11(2):37 and other electricity generation technologies due to effects of
Kirshen P, Ruth M, Anderson W (2008) Interdependencies of urban climate change. ECORYS, Netherlands
climate change impacts and adaptation strategies: a case study of Rajani B, Kleiner Y (2001) Comprehensive review of structural
metropolitan Boston USA. Clim Chang 86(1–2):105–122 deterioration of water mains: physically based models. Urban
Kleiner Y, Rajani B (2001) Comprehensive review of structural Water 3(3):151–164
deterioration of water mains: statistical models. Urban Water Rajani B, Tesfamariam S (2004) Uncoupled axial flexural and
3(3):131–150 circumferential pipe-soil interaction analyses of partially sup-
Koch H, Vogele S (2009) Dynamic modeling of water demand water ported jointed water mains. Can Geotech J 41(6):997–1010
availability and adaptation strategies for power plants to global Rijkswaterstaat (2009) Jaarverslag 2008 en plan 2009–2010: Bureau
change. Ecol Econ 68(7):2031–2039 Veiligheidsbeambte
Koetse MJ, Rietveld P (2009) The impact of climate change and Rosato V, Issacharoff L, Tiriticco F, Meloni S, Porcellinis S, Setola R
weather on transport: an overview of empirical findings. Transp (2008) Modelling interdependent infrastructures using interact-
Res Part D 14(3):205–221 ing dynamical models. Int J Crit Infrastruct 4(1/2):63–79
Oostroom H, Annema JA, Kolkman, J (2008) Effecten van klimaat- Rosmuller N, Lievit M, Snelder M, Tonnaeer C (2011) Incidentman-
verandering op verkeer en vervoer Implicaties voor beleid. agement en stremmingskosten: vergelijking van een weg- sche-
Kennisinstituut voor Mobiliteitsbeleid epvaart- en spoorincident. Stichting Platform Transportveiligheid
Laird J, Nellthorp J, Mackie PJ (2005) Network effects and total Rothstein B, Parey S (2011) Impacts of and adaptation to climate
economic impact in transport appraisal. Transp Policy change in the electricity sector in Germany and France. In: Ford
12(6):537–544 JD, Berrang-Ford L (eds) Climate change adaptation in devel-
Larsen P, Goldsmith S, Smith O, Wilson ML, Strzepek K, Chinowsky oped nations. Springer, New York, pp 231–241
P, Saylor B (2007) Estimating future costs for Alaska public Rothstein B, Schroedter-Homscheidt M, Haefner C, Bernhardt S,
infrastructure at risk from climate change. Glob Environ Chang Mimler S (2008) Impacts of climate change on the electricity
18(3):442–457 sector and possible adaptation measures. In: Economics and
Linnerud K, Mideksa TK, Eskeland GS (2011) The impact of climate Management of Climate Change. New York: Springer,
change on nuclear power supply. Energy J 32(1):49–168 pp 231–241
Lobato E, Rouco L, Gomez T, Echavarren FM, Navarrete MI, Simon HA (1962) The Architecture of Complexity. In: Proceedings of
Casanova R, Lopez G (2004) Preventive analysis and solution of the American Philosophical Society, pp 467–482
overloads in the Spanish electricity market. ElectrPower Syst Simon HA (1973) The organization of complex systems. In: Pattee
Res 68(3):185–192 HH (ed) Hierarchy theory—the challenge of complex systems.
Margerum RD (2011) Beyond consensus: improving collaborative Goerge Braziller, New York, pp 1–27
planning and management. MIT Press, Cambridge Snelder M (2010) Designing Robust Road Networks: a general design
Meadows DH (1999) Leverage points: places to intervene in a system method applied to the Netherlands. T2010/10 TRAIL Research
Hartland VT. The Sustainability Institute School, the Netherlands
Mideksa TK, Kallbekken S (2010) The impact of climate change on Sterman J (2000) Business dynamics: systems thinking and modeling
the electricity market—a review. Energy Policy 38(7): for a complex world. Irwin/McGraw-Hill, Boston
3579–3585 Stern N (2007) The economics of climate change. Cambridge
Mili L (2011) Making the concepts of robustness resilience and University Press, Cambridge
sustainability useful tools for power system planning operation Susskind L, Crump L (2009) Multiparty negotiations. Sage, Thousand
and control. 4th International Symposium on Resilient Control Oaks
Systems SWOV (2009) De invloed van het weer op de verkeersveiligheid (the
National Research Council (2009) Informing decisions in a changing influence of weather on traffic safety). http://www.swov.nl/rapport/
climate. The National Academies Press, Washington, DC Factsheets/NL/Factsheet_Invloed_van_het_weer.pdf. Accessed
Newport R (1981) Factors influencing the occurrence of bursts in iron July 2012
water mains. Water Supply Manag 3:274–278 Tennet (2010) Overhead line monitoring. http://wwwtennetorg/english/
Niemczynowicz J (1999) Urban hydrology and water management— images/100552%20Overhead%20line%20monitoring_tcm43-19467
present and future challenges. Urban Water 1(1):1–14 pdf. Accessed May 2012

123
A framework for supporting governance 931

Transportation Research Board (2008) Potential impacts of climate systems. Water Science & Technology: Water Supply—
change on US transportation. The National Academies Press, WSTWS 9(6): 635–641
Washington, DC Webb J (2011) Making climate change governable: the case of the
Tzimas E (2011) Sustainable or not? Impacts and Uncertainties of UK climate change risk assessment and adaptation. Plan Sci
Low-carbon Technologies on Water’ presentation at AAAS Public Policy 38(4):279–292
February 2011 European Commission—Joint Research Centre Wilbanks TJ, Bhatt V, Bilello DE, Bull SR, Ekmann J, Horak WC,
Institute for Energy mimeo Huang YJ, Levine MD, Sale MJ, Schmalzer DK, Scott MJ
Van Daal K, Blokker M, Holzhaus P, Hendrikx D (2008) Storingen (2008) Effects of climate change on energy production and use
als early warning system voor conditie van leidingnetten in the United States. US Climate Change Science Program,
H2O(19):95–97 Washington, DC
Van Vliet MTH, Yearsley JR, Ludwig F, Vgele S, Lettenmaier DP, Williams BK, Szaro RC, Shapiro CD (2009) Adaptive management:
Kabat P (2012) Vulnerability of US and European electricity the US department of the interior technical guide. US Depart-
supply to climate change. Nat Clim Chang 2(9):676–681 ment of the Interior, Washington, DC
Vreeburg JHG, Blokker EJM, Horst P, Van Dijk JC (2009) Velocity-
based self-cleaning residential drinking water distribution

123