Introduction

There is wide recognition that modern society depends on goods and services provided by a set of complex systems typically referred to as critical infrastructures. These systems are often referred to as critical since they are essential for maintaining and sustaining public well-being, safety, and economic prosperity [26, 42, 55, 73]. The domain of critical infrastructures revolves around chemicals, commercial facilities, communications, critical manufacturing, dams, defense industrial bases, emergency services, energy, financial, services, food and agriculture, government facilities, healthcare and public health, information technology, nuclear reactors, materials, and waste water systems [67]. Lately, there is increasing interest in the energy sector with respect to the critical importance of Smart Grids as a critical infrastructure [11, 15, 54, 57, 72, 75]. Arguably, Smart Grids, similar to all critical infrastructures, operate under conditions of uncertainty with respect to natural events such as earthquakes and hurricanes as well as man-made events such as acts of sabotage and cyber-threats [42, 80, 86, 92]. Moreover, Smart Grids must be designed, operate, and evolve in a difficult context. This context is marked by elements of: (1) ambiguity associated with an increasing lack of clarity and situational understanding, complexity associated with large numbers of richly and dynamically interacting systems and subsystems with behavior difficult to predict, (2) emergence with respect to the inability to deduce behavior, structure, or performance from constituent elements, and (3) interdependency associated with mutual influence among different complex systems through which the state of a system influences, and is influenced by, the state of other interconnected systems [12, 51].

Against this backdrop, current research tends to focus on the potential benefits of Smart Grids [23] as well as issues and risks in implementation such as new cyber-threats and vulnerabilities [7, 13, 14, 36, 37]. Moreover, and perhaps due to the nascent nature of this topic, researchers are still debating the definitions of Smart Grids [21, 40, 70, 78] as well as focusing on particular aspects and parts/elements of Smart Grids, including design for next-generation control centers [91], optimizing distributed power systems [75] effects of plug-in-hybrid-electric vehicles [33], security issues [4, 6, 8, 62, 64], Smart Meters [90], standards and best practices [32, 87], and classification of threats [4, 15, 54] among others. However, there is still a scarcity of literature discussing quantitative methods that could be used in support of risk quantification for Smart Grids.

The idea of risk quantification for Smart Grids is not new. Concepts of probability of occurrence of an event and its consequences have been adapted for Smart Grids [11, 32, 40, 57–59, 62, 72, 90]. However, adapting traditional risk formulation without accounting for other relevant measures is limiting in analyzing Smart Grids. Moreover, current literature could be considered atomistic since there is a tendency to focus on specific elements such as smart metering systems and integration of distributed power generation of the Smart Grid, without consideration of Smart Grids as a totality. Thus, there is a gap in the literature for developing more robust formulations of risk related to more holistic considerations for integrated Smart Grid systems. This research attempts to address this gap by exploring a robust set of measures (and their properties) that could be used for more holistic examination of Smart Grids. The purpose of this paper is to propose and develop an alternative framework that could be used to explore the ‘criticality’ of Smart Grids. For purposes of this research, the term criticality is related to the importance of a Smart Grid to public well-being.

The paper is organized around three primary development thrusts to support the purpose of the research. First, we describe Smart Grids in terms of the present domain and major characteristics that delineate the domain. The aim of this section is to articulate the complexities involved in developing, implementing, and evolving Smart Grids. Special emphasis is placed on the more holistic view of Smart Grids as an integrated system that includes technologies, information (availability, accessibility, utility), human and social influences, organizational and managerial supporting arrangements, and political (policy) constraints as well as facilitation considerations. Second, the concept of risk is explored. Specifically, the literature is reviewed with respect to risk and factors commonly used in current quantification efforts related to risk for Smart Grids. Third, we provide a preliminary extended set of measures that could be used in addressing criticality of Smart Grids. This set of measures and their properties is developed by contrasting current factors with previous research of criticality-based measures. This research concludes with proposed future research directions based on the current investigation implications.

Smart Grid Characteristics and Landscape

The topic of Smart Grid is relatively new and as such there is no one widely accepted definition [40, 70]. Thus, it is necessary to explore the concept of Smart Grid to develop a foundational perspective before delving into the concept of risk for Smart Grids. At a fundamental level, a Smart Grid can be considered “an upgrade to the current electrical power grid” ([8], p. 24). Consequently, a Smart Grid is expected to meet current needs while offering significantly higher capabilities that are intended to meet ever changing societal demands of the 21 st century and beyond [54]. These social demands are highlighted by the need for reliable, resilient, scalable, manageable, and environmentally friendly energy generation, transmission, and distribution systems that also embody concepts of interoperability, cost effectiveness, and intelligence [6, 23, 31, 54]. Unfortunately, there is no one consistent perspective of what a Smart Grid entails. Table 1 is provided to illustrate the varying representative perspectives of Smart Grids.

Table 1 A representative set of perspectives on smart grids
Full size table

Although, there is no one accepted perspective of Smart Grids, the selected set of perspectives begins to offer insight into essential aspects, components and characteristics of Smart Grids. A common theme of transforming the structure of electrical energy generation, delivery and consumption with an increasing emphasis on information and technology and interests of the consumer, appear to be driving the evolving paradigm of Smart Grids. The Institute of Electrical and Electronics Engineers (IEEE) definition, “integration of power, communications, and information technologies for an improved electric power infrastructure serving loads while providing for an ongoing evolution of end-use applications” ([38], p. 3), appears to capture most perspectives. To further expound on these representative perspectives, we now turn attention to the common basic components and characteristics of Smart Grids.

Similar to differing Smart Grid perspectives, there are also different perspectives on components (elements) that constitute Smart Grids. Baumeister [8], from a cybersecurity perspective, suggests that there are five categories of major themes for Smart Grids. Table 2 presents these categories along with their typical associated elements. These themes are directly related to each being a “component of the Smart Grid” ([8], p. 6) from the security perspective of Smart Grids.

Table 2 Baumeister’s [8] categories and components of the smart grid
Full size table

Describing Smart Grids in terms of security domain appears as a dominant theme in literature. This might be attributed to increased coupling of information in the energy sector which has created new vulnerabilities [61]. These new vulnerabilities are especially inclusive of threats of the cyber-kind [13, 78, 86, 87].

An alternative approach for describing Smart Grids is provided through the lens of architectural representations. A block diagram is provided by Balaji and Ram [6] to illustrate Smart Grid as a “vast network comprising utilities and customers who are linked by the power transmission as well as communication infrastructure. The other entities in the network are involved in providing value added services for improving efficiency and facilitation of buying and selling of power driven by supply demand dynamics” ([6], p. 2903). A hierarchical architecture model has also been suggested by Moslehi and Kumar [63]. Their approach is based on the need for “harnessing modern communication and information technologies to enable an IT [Information technology] infrastructure that provides gridwide coordinated monitoring and control capabilities” and as such, it is mainly focused on “operating concerns in categories such as performance enhancement, equipment limits, operating limits, system protection, and rapid recovery” ([63], p. 60) with an emphasis on functional tasks of the elements comprising a Smart Grid. Yet another model of a Smart Grid is suggested by Komninos et al. [54] in the form of a multi-layered conceptual model that illustrates three major sections of a Smart Grid as well as its parts and their interactions. These representations provide a means by which a typical Smart Grid can be viewed as a complex of interrelated parts and elements [10, 32]. Consequently, these views are compatible with contemporary research trends of focusing on elements, their interactions in a grid, and exchange of information [32, 62, 72].

Perhaps a more comprehensive view of the Smart Grid is provided by IEEE’s Standard 2030-2011 [38]. This standard articulates major entities and functions of a Smart Grid that aligns with the National Institute of Standards and Technology [NIST] framework for Smart Grids. Figure 1 is adapted from NIST [65] to depict the seven domains of a Smart Grid. The solid blue lines indicate the secure information and communication flows. The red dotted lines represent electricity flows.

Fig. 1
figure 1

A conceptual model of a smart grid

Full size image

Each of the seven ‘domains’ can contain a number of interrelated complex systems along with logical interfaces (i.e., access points) in which information can enter/exit a domain [38, 65]. Table 3 depicts the seven domains of a Smart Grid as well as entities commonly associated with those domains. While this table does not place emphasis on the interfaces among the different domains of a Smart Grid, it forms the basis for suggesting that each domain could be viewed as a complex system. Guckenheimer and Ottino’s [30] four distinctive properties of a complex system (i.e., many interacting parts, emergent behavior, adaptation and change, systems uncertainty) appear present for Smart Grids. For example, Advanced Metering Infrastructure, which is a building block for the customer domain, is described as a complex set of interrelated elements [4, 17, 62].

Table 3 Domains and entities of a smart grid
Full size table

Arguably, when it comes to describing Smart Grids, none of these perspectives are incorrect. In fact, these perspectives are all necessary to set the basis for developing best practices for designing, maintaining, and realizing the premises of Smart Grids [32, 70, 84]. In fact, the perspectives also serve as the basis for creating measures and indicators that are instrumental in assessing performance of Smart Grids [78]. Inevitably, these perspectives can be used to enhance our understanding of the logic of various domains and their interrelations in Smart Grids [62]. Beyond the different articulations of Smart Grids, and perhaps more importantly, is the fact that there appears to be a set common themes that describe general ‘characteristics’ of Smart Grids [6, 14, 63, 68, 78]. Table 4 provides a set of common set of ‘characteristics’ of Smart Grids that are drawn from pertinent literature.

Table 4 Unifying characteristics of smart grids
Full size table

The discussion regarding components and the unifying ‘characteristics’ of Smart Grids were purposefully selected to illustrate three important points. First, the topic of Smart Grid is still in its infancy and therefore should be expected to be loosely bounded and harbor a degree of diverging and sometimes conflicting perspectives. Having diverging perspectives is not troubling. Rather, as [48] suggest, different perspectives put forward “show the potential sources of divergence in the development of the [Smart Grid] field …[with] Each perspective brings[ing] a logic which provides its own internal validation to the community which produces and consumes the perspective” ([48], p. 240). Second, the field of Smart Grids can be identified as existing within the domain of critical infrastructure which “addresses elements of assessment, remediation, indications and warnings, mitigation, response, and reconstruction pertaining to hazards, risks, and threats from natural and manmade events affecting public well-being, public safety, economic vitality, and security” ([25], p. 194). Increasing concerns about frequency of occurrence of risk events, such as breaches, as well as their potential effects on public well-being, highlights the relative importance of Smart Grids as it relates to public well-being, including considerations of health, security, and economic impact [11, 47, 62, 90]. Third, the operating landscape/environment for Smart Grids is characteristically complex, involving a range of socio and technical issues [92]. The articulation of the current state of the Smart Grids problem domain can be characterized consistent with earlier works [45, 46, 48–51] and the notion of ‘messes’ by [1] as well as ‘wicked problems’ by [74]. Table 5 provide articulates characteristics of a landscape from which Smart Grids are projected to operate. This operating landscape suggests a need for robust analysis methods in all aspects of realization of Smart Grids.

Table 5 The operating landscape for smart grids and implications
Full size table

One key aspect of realization of Smart Grids is risk. There is a growing support suggesting the need to address ‘risk’ related to Smart Grids [11, 15, 32, 33, 40, 54, 57, 62, 72, 75, 90, 91]. The following section provides an initial exploration into the concept of risk for Smart Grids as well as its quantification.

The Concept of Risk in Smart Grids

There is no one widely accepted definition of the term ‘risk.’ However, risk is typically defined in terms of probability of occurrence of an event and the magnitude of the resulting consequences [5, 24]. The vast literature on this topic also suggests that elements of event sequence and probabilities [71], technical factors in a system life cycle [39], probabilities of unknown outcomes and uncertainties [28], uncertainty [35, 53], perception of risk [88], mental constructs of risk [26], and ‘unknown unknowns’ [69] are also essential considerations related to risk quantification. Most recently, the concept of interdependency is increasingly incorporated into risk quantification [47, 73, 79]. Regardless of subtle and wide ranging distinctions related to the risk, a general consensus is that occurrence of a risk event can cause undesirable effects related to such issues as cost, schedule, and or technical performance of a system [18].

There is no shortage of risk events that can affect performance of Smart Grids. These range from natural events such as extreme weather conditions [57, 75] to man-made acts such as cyber-threats [4, 21]. The increasing frequency of such events coupled with their potential negative effects on public well-being suggests a need for development of risk-related approaches that could be used in understanding such emerging risks as well as aid in decision-making processes to mitigate their impact and/or prevent their occurrence altogether. However, we contend that current literature related to risk in relationship to Smart Grids suffers from two primary deficiencies: (1) it accounts for a limited set of traditional factors for quantification of risk and (2) it is atomistic in analysis for Smart Grids since it focuses on specific domains and elements of Smart Grids (e.g. transmission). These issues form the basis for remainder of this article as well as development of an extended set of measures that could be used in a more systemic analysis to aid in understanding and designing Smart Grids.

First, it is essential to recognize that the traditional risk formulation has been adapted for application to Smart Grids. For example, [59] suggests that an overload risk assessment for a transmission line can be drawn from probability overload at a given line and the severity of the overload on the system. Rocchetta et al. [75] have also developed a simulation-based risk-cost optimization framework that accounts for high wind, solar irradiation, and lightning as major issues affecting failure rates in the overhead distribution lines of Smart Grids. The summation of probability of undesired events and the severity of the related consequences are used as the primary factors for risk articulation related to Smart Grids. Corresponding contingency frequencies related to unexpected loss of one or more elements (e.g. distribution line, transformer, etc.) and overload are used for risk estimation. Specifically, Rocchetta et al. [75] used a Monte Carlo simulation approach with a continuous Weibull distribution to illustrate the importance of integrating distributed power systems into a Smart Grid environment to counter the effects of extreme weather on availability of electricity. These approaches are similar to those of [40] and [91] with both using traditional factors of probability and consequence, although in different context, architecture of a Smart Grid for the former and Smart Control Center for the later.

Undoubtedly, electric, hybrid electric, and plug-in hybrid electric vehicles present a desirable potential for substantial impact on pollution, climate change, and energy utilization. However, and as indicated by Hashemi-Dezaki et al. [33], increased and especially unmanaged charging of these vehicles “may adversely [the] affect electric distribution system” ([33], p. 262). Hashemi-Dezaki and his colleagues [33] suggest that implementing managed charging with a schedule for charging plug-in hybrid electric vehicles is beneficial as it does not compromise the reliability of a Smart Grid. Similar to [75], Hashemi-Dezaki et al. [33] also uses a Monte Carlo simulation. However, risk is directly tied to reliability measures of mean time to failure (MTTF) and mean time to repair (MTTR). These examples point to the need for holistic consideration of Smart Grids to better capture unintended consequences which may accrue as the system operates.

Beyond measures articulated above, literature also indicates a unique set of factors that could be used in association with analysis of Smart Grids. For example, [62] suggests that risk analysis for a utility system could involve the goal of the adversary (i.e., motivation for attacking a business), threat agent availability, potential threat vectors, exposure, target attractiveness, and impact of attack. These measures, according to [62] are instrumental in identification of vulnerability, prioritization of the threats, and development of countermeasures. Table 6 is a summary of literature depicting risk-related measures in different areas of Smart Grids.

Table 6 A synthesis of literature positioning in different aspects of risk for smart grids
Full size table

This section was developed to illustrate how risk is currently being addressed in the Smart Grid literature as well as the implications for further development. Authors draw two primary conclusions based on this literature. First, it is evident that contemporary literature focuses on risk separately in the specific domains and elements associated with Smart Grids. This would then suggest that approaches for risk quantification would necessarily be expected to vary from domain to domain within Smart Grids. For example, the approach for transmission lines [59] significantly varies from a security-related approach [62]. It might be reasonably expected to have different approaches in the different domains of Smart Grids since certain factors are domain-specific. However, this focus on risks related to the constituent domains offers limited utility to those who might be involved at an integrated level of Smart Grids, which exist beyond individual domains. At the higher (systems) level of Smart Grids, risks cannot be assumed to be aggregates of mutually exclusive and independent risks of the constituent domain risks. Thus, risks at the Smart Grid exist at a different logical level than those of the constituent domains and cannot be simple inferred from domain level risks. An analysis based on simple extrapolation of constituent domain risks are tenuous at best and outright wrong at worst. If the objective is to analyze Smart Grids at a system level, we must look beyond simple aggregation of risks from constituent domains. Second, traditional risk formulation of probability and consequence is prevalent across the literature and in the individual constituent domains for Smart Grids and offers a good starting point to rethinking analysis of the higher logical (systemic) level for Smart Grids. More advanced, and arguable more appropriate, risk literature also points to consideration of a more robust set of factors that might be useful quantifying risk for Smart Grids (e.g. dependency, interdependency, resilience). It is from this perspective that we propose developing an extended set of factors that could be used in more holistic analysis of Smart Grids.

Holistic Risk Formulation for Smart Grids

The need for holistic approaches to risk analysis for Smart Grids is not new [32, 40]. It has long been recognized that the evolving nature of threats coupled with the ‘E + I’ (i.e., energy and information) paradigm [26] have transformed the thinking from the traditionally isolated systems perspective into a more “highly interconnected and interdependent system of local and wide area information and communication systems” ([72], p. 276) perspective. This means consideration of interactions and interdependencies among different domains of Smart Grids (i.e., generation, transmissions, distribution, customers, markets, operations, and service providers) as well as the other systems in the environment [40]. Under this emerging paradigm, Ray and his colleagues suggest the development of a unified risk model that considers “interconnections of domains...variety of dynamic and structural interactions” for Smart Grids ([72], p. 281). In light of these insights, there is an increased call to rethink how risk is formulated [42, 43, 47] with respect to the complexities that are endemic to modern systems and higher level domains such as Smart Grids.

In this section, we develop an extended list of factors that can be used for analysis of Smart Grids. Although it is not presented as the definitive listing of factors, it is offered as a first articulation for moving beyond the narrower conceptions of risks in Smart Grids. As such, the factors can, should, and will evolve with further development, insights, and applications. As a first step in this exploration, we examine a set of extended factors identified from the Smart Grid literature. A review of existing factors suggests that a total 10 factors from the literature (i.e., attractiveness of a target, availability of threat agents, availability of attack routes, ineffective protection measures, exposure, reliability, tolerance of stakeholders, total number of lines, total number of nodes, and vulnerability). These factors are in additional to the two traditional factors of probability and consequence that are used in different risk-related approaches for Smart Grids. Many of these factors are terms or phrases that have an exact or nearly the same meaning used in different contexts – that is they are synonyms. For example, Bologna et al. ([11], p. 6) refers to “potential attack paths” to suggest a route that could be used to attack a system. On the other hand, McBride and McGee ([62], p. 97) use the words “potential threat vectors” to suggest potential ‘channels’ such the “Internet, wireless access points, the enterprise intranet, mobile devices (including USB devices), remote endpoints (including meters), the supply chain, and the company’s own systems development organization” that could be used in attacking a system. Clearly, the different phrases address congruent issues and are referring to availability of routes/paths that can be used to exploit a system - in this case, a Smart Grid. The same logic is used to combine concepts of ease of attack, effectiveness of countermeasures, and lack of security measures. Table 7 summarizes a synthesis of similar terms into seemingly unique factors that can be applied at the level of Smart Grids for risk quantification.

Table 7 A set of factors for smart grid risk quantification
Full size table

Arguably, Katina and Hester [42] have developed a comprehensive set of factors that can be used to determine ‘criticality’ of infrastructure systems [85]. To this end, we can compare contemporary Smart Grid risk factors to criticality-based factors to develop a comprehensive set of factors to support analysis of Smart Grids. Katina and Hester [42], researchers with the National Centers for System of Systems Engineering, in their attempt to create a generalizable and transportable method for prioritizing critical infrastructures, postulated that current methodologies are sector-specific approaches and/or based on regional factors. They proposed a four-tuple of ‘criticality’ factors of levels of resiliency, level of interdependency, level of dependency along with infrastructure risk Footnote 1 as fundamental to ranking and prioritization of infrastructures regardless of sector or region. The term ‘criticality’ in this sense relates to the importance of an infrastructure (e.g., a Smart Grid) to public well-being. Each of the four factors of criticality are associated with a set of properties that could be used for measuring each factor and contributes to a set of higher level criticality measures for a system. Figure 2 is drawn to capture the essence of Katina and Hester’s [42] four-tuple criticality-based measures.

Fig. 2
figure 2

Areas of a criticality-based approach

Full size image

A mapping of the 10 seemingly unique factors from the Smart Grid literature into the criticality-based measures reveals a number of issues. First, several factors from the Smart Grid literature can be merged into single factors of the criticality-based approach. For example, total number of lines [F8] and total number of nodes [F9] are essentially addressing the effects of having a great number of interconnected systems. Nodes represent systems and lines represent the means by which such systems are interconnected. Failure in such a system, from a criticality-based analysis approach, is described in terms of number lines that fail after an attack, which in turn affects nodes in a Smart Grid. This issue is addressed in the interdependency criticality-factor since it includes interconnectedness of a system [42]. F4 is also combinable with F6 inasmuch as F6 is not possible if a system has ineffective protection measures. Table 8 provides a mapping of contemporary risk factors for Smart Grid risk (synthesized from literature) to those proposed by [42].

Table 8 Mapping contemporary Smart Grid risk-related factors into criticality-based model for infrastructure systems
Full size table

Second, there is a gap in how risk for Smart Grids is addressed. As indicated in Table 8, there are a number of properties that could be associated with one another. For example, ‘community awareness’ is associated to ‘measure of dependency’, ‘external relationship’ is associated to ‘measure of interdependency’, ‘system protective’ characteristics is associated to ‘resiliency’ and ‘environmental factors’ associated to ‘risk.’ These properties, in addition to probability and consequence, can be used to inform a more robust and holistic analysis for Smart Grids. While these properties are not presented as exhaustive, they offer a more extensive set of ‘metrics’ for a deeper and more rigorous analysis of Smart Grids.

Finally, applying traditional measures of probability and consequence to a specific domain of a Smart Grid offers only a partial view of the landscape within which Smart Grids operate. Subsequently, there is a need to consider the interrelationships among the seven different domains of Smart Grids. Arguably, these relationships could be explored in terms of dependency, where functioning of a given domain (e.g., distribution) is dependent on another domain (e.g., transmission). The interdependency measure recognizes that each domain (e.g., customer) influences and is also influenced by the remainder of the domains of a Smart Grid. The resiliency measure initiates the discussion about designing Smart Grids that can withstand or rapidly recover from threats and hazards. These measures are in addition to the traditional considerations of probability and consequence associated with risk. Together, the four measures of the criticality-based approach (CBA) for analysis of Smart Grids account for several properties that could be instrumental in design, development, and analysis of Smart Grids. An operand for the proposed approach is provided as:

$$\begin{array}{@{}rcl@{}} {Cr}_{SmartGrid}&=&f({Level}_{Dependency}{Level}_{Interdependency}\\&&\ominus {Level}_{Resiliency}\ominus {Level}_{Risk}) \end{array} $$

Each level of measure, which for simplicity might range 0 to 10, could be assessed based on the proposed properties as suggested below to offer information and insights on the state of a Smart Grid. The combination of these measures could then be used in a radar chart for the analysis of a Smart Grid as indicated in Fig. 3. Each spoke of the radar chart represents one of the four measures. In Fig. 3, the observation of a 2 for a measure of risk suggests a low risk level in a given scenario.

Fig. 3
figure 3

Criticality measures for Smart Grid and their representation in a radar diagram

Full size image

The proposed four-tuple measures certainly contribute to current research in different ways. At the framework level, current instantiations of risk-based frameworks have a “set of optimal steps [phases] that can be used identify, evaluate and control risk to mitigate potential negative effects in Smart Grid[s]” ([90], p. 89). Typically, these phases include risk identification, risk characterization, risk evaluation, risk mitigation planning, risk management, risk communication, and monitoring and review process at the conclusion. The proposed approach complements risk-based frameworks for Smart Grids in identifying potential issues that could affect performance as well as areas that could be in need of attention. For example, the properties associated with dependence such economic importance could enable the analyst to consider where the provision of goods and services of a Smart Grid are economically feasible. In the consideration of economic feasibility, the analyst might deliberate the role of malicious, technical, and/or natural hazards affecting the system. Therefore, this research offers a different lens through which policy-makers, Smart Grid owners, and operators might analyze Smart Grids beyond the traditional perspective of risk limited to probability and consequence. Also, observations of the different levels (continuous or incremental) of the properties might offer insights into the state of the Smart Grid such that indicators supporting more robust changes could be detected and examined.

Conclusions and Future Directions

Smart Grids, to meet the challenges and satisfy the needs of the context from which they are derived, will fundamentally be required to address a variety of issues present in their current operating landscape. Arguably, the operating landscape for Smart Grids requires that we rethink how to address risk to truly realize the full potential and contributions sought for Smart Grids. A strict view of risk that considers only probability of occurrence of an event that could halt Smart Grid operations and consequences of such an event on public well-being, offers limited utility for application to the complex nature of Smart Grids. Such a limited approach is likely to produce an overly narrow and short-term view of risk for practitioners who must contend with a spectrum of issues that could affect performance of Smart Grids. This paper proposes an approach: criticality-based approach (CBA), for the analysis of Smart Grids with four measures: dependency, interdependency, resiliency, and risk (inclusive of traditional probability of occurrence and consequences). Each category measurement involves a set of properties that could be used in design, analysis, and evolution of Smart Grids as well as the development of countermeasures for issues associated with performance of Smart Grids.

While a CBA for analysis of Smart Grids is a necessary step in a robust analysis, much research remains for realization and operationalizing this approach. A primary area for development remains how to measure the different properties that contribute to the different measures associated to CBA centered on dependency, interdependency, resiliency, and risk. A starting point should certainly involve on a review of how the elements of the four-tuple are currently measured. For example, a measure of interdependency has been proposed in literature [47, 76]. These could be adapted for Smart Grid research as well as linguistic measures (i.e., low, medium, and high) which could then be translated into numerical values [15, 27, 72]. This becomes a starting point for applications and quantification of the proposed approach for analyzing Smart Grids. Two major contributions would be: (1) the ability to compare and contrast the states of different Smart Grids and impacts of improvement initiatives and (2) establishing a baseline against which the development and improvement of a Smart Grid could be more rigorously measured.

A Smart Grid is part of the energy sector and thus related to critical infrastructures enabling production of goods and services essential for public well-being. Public well-being is intrinsically tied to measures of the CBA in Smart Grids analysis. However, there are no known well-articulated indicators or tools for measuring public well-being in relationship to Smart Grids. In response to this gap, such indicators could be developed and explicitly attributed to goods and services provided as a result of Smart Grids. This might provide a basis for relating each of the measures of the proposed approach, as well as their properties, to public well-being. This ‘measurable’ relationship could then form the basis for more informed decisions-making concerning allocation of scarce resources, prioritization of Smart Grid development, exploration of potential scenarios, and establishment of the level of tolerance for different issues affecting Smart Grids. In accordance with the latest reports (see [10]), such tools have to be developed and ‘lab tested’ to ensure operability in the real world.