After conducting an overview of the 222 papers, content analysis was used as a means for determining the primary streams of this IR-related literature. The statements, data, figures, and tables existing in these studies were analyzed in detail during this process. Finally, five streams of the infrastructure resilience research were identified, as shown in
Table 6. These research streams include the assessment of infrastructure resilience, the improvement of infrastructure resilience, conceptualizing infrastructure resilience from various perspectives, factors influencing infrastructure resilience, and the prediction of infrastructure resilience.
Table 6 reflects the number of studies on each stream in the field of IR. Among 222 identified pieces of literature, over half of the total focused on the assessment of IR, with a proportion of 54.1%. The number of research on improving IR occupies the second position, taking up 27.0%. Literature on decomposing IR and Influencing IR are, respectively, responsible for 10.4% and 6.3%. Additionally, the number of research on predicting IR is the least, only accounting for 2.2%. The major research efforts carried out in these streams are elaborated on in the following sections.
4.1. Assessment of Infrastructure Resilience
Assessing infrastructure resilience is critical to determining preventive measures to mitigate the consequences caused by various disruptive events [
29]. A large number of researchers have assessed IR responding to disruptions by adopting different analytical tools. For example, Levenberg et al. [
30] quantified and assessed the resilience of networked pavement infrastructure by modeling a set of possible network performance scenarios in a destructive meteorological scenario with a known probability of occurrence; each scenario was defined according to the severity and type of damage (climate or geology, operation, natural deterioration, and terrorism), as well as current weather conditions, temperature, precipitation, and visibility conditions. Sun, Stojadinovic, and Sansavini [
31] put forward an agent-based modeling framework for the seismic resilience assessment of integrated civil infrastructure systems under the scenario of an earthquake. Barabadi et al. [
32] assessed the resilience of health infrastructure before and after COVID-19. Xu, Cong, and Proverbs [
33] proposed a multidimensional evaluation index system for assessing the resilient capacity of infrastructure systems to cope with the extreme weather in Wuhan.
Quantifying resilience is a challenging problem [
34]. Bruneau et al. [
16] proposed a comprehensive assessment framework of resilience, pointing out that 11 aspects need to be considered when assessing resilience (
Figure 4), including 4 dimensions (i.e., technical, organizational, social, and economic), 4 basic properties (i.e., robustness, rapidity, redundancy, and resourcefulness), as well as 3 outcomes (i.e., more reliable, faster recovery, lower consequences). Huck, Monstadt, and Driessen [
35] applied the 4R concept integrating robustness, rapidity, redundancy, and resourcefulness in assessing the resilience of power infrastructure in order to establish disaster management. Toroghi and Thomas [
36] used a framework including five metrics (robustness, resourcefulness, redundancy, rapidity, and readjust-ability), which assesses the resilience of electric infrastructure systems using the 4R concept and the readjust-ability.
Robustness is one of the most important properties of resilience, referring to the ability of the system to survive under abnormal and hazardous conditions [
37]. When one or more of the capacities of the interconnected system are exhausted, a resilience failure event will occur, which firstly relates to the system’s robustness [
38]. A variety of researchers assessed the resilience of infrastructure in terms of robustness [
39,
40,
41]. Homayounfar et al. [
40] used a stylized dynamical model for assessing the connection between resilience and robustness of coupled infrastructure systems, stating that robustness is a component of resilience; this model simulates how the robustness and resilience of critical infrastructure systems respond to shocks in state variables change with parameters. Huizar et al. [
41] used robustness as a metric for quantifying the resilience of supply and demand of water infrastructure systems, providing a reference for water infrastructure planning and design.
Redundancy can enhance the reliability of a system through the duplication of its critical components or functions, which is one of the metrics measuring the system’s resilience [
42]. Perz et al. [
43] adopted a three-country border in the southwestern Amazon that was integrated by a highway as a case study, and they used rural survey data to assess the relationship between the connectivity of community and town and socioecological resilience, exploring the effects of resilience and connectivity on infrastructure. Capacci and Biondini [
44] proposed a probabilistic framework for assessing the life-cycle seismic resilience of infrastructure networks integrating the aging bridges and transportation roads, thus exploring the approach to strengthening the redundancy of infrastructure networks.
Rapidity refers to the capacity of a system to meet priorities and achieve goals in time after disruptions to control enormous losses, which indicates minimizing the time required to recover to full system operations and productivity [
45]. A number of researchers evaluated IR using this property [
46,
47,
48,
49]. Cimellaro et al. [
46] presented recovery curves based on the time series recorded from March 11 to April 26 during the 2011 Tohoku Earthquake in Japan to assess the resilience of physical infrastructure; the recovery curves, respectively, indicate the restoration ratio between the number of households without service and the total number of households in terms of three categories of lifelines along with the time series, including power delivery, water supply, and city gas delivery. Argyroudis et al. [
48] adopted a multi-hazard assessment framework to quantify the resilience of critical infrastructure systems; considering the impacts of cascading hazards, where the subsequent hazard is triggered by the initial hazard simultaneously or within a short period, and the restoration commences after the completion of the multiple hazard sequences, this framework measures the vulnerability of the assets to risks and the rapidity of the restoration in the scenario of cascading hazards.
As a property of measuring resilience, resourcefulness is the ability of a system to allocate resources rationally to minimize the impacts of hazards and improve its performance [
50]. Vadali et al. [
51] proposed the optimal approach to maintaining infrastructure resilience by using a bi-national dynamic traffic assignment model; this model provides a timely insight into the daily travel impact and economic cost of an unexpected disruption to the ports-of-entry infrastructure. Lau et al. [
52] identified the most efficient approach to meeting the demand shortage of critical infrastructure and maintaining the resilience of the power grid by establishing a grid optimization model, which consisted of the low-level (micro-grid) and mid-level voltage grid components in urban power grids for disaster recovery.
Compared with the research assessing IR from a perspective of a single property, the majority of reviewed papers developed and quantified the resilience of infrastructure systems based on comprehensive properties. For example, Shafieezadeh and Burden [
53] proposed a probabilistic framework for the scenario-based resilience assessment of infrastructure systems; this method considered the uncertainty in the process, including the correlation of seismic intensity measurement, vulnerability assessment of structural components, estimation of maintenance demand, maintenance process, and service demand. Zhu et al. [
54] used a flexible assessment framework that comprised eight metrics adapted from existing research: vulnerability, expectations, redundancy, adaptability, rapidity, intelligence, cross-scale interaction, and learning culture, to assess the resilience of the power and water infrastructure systems in Kathmandu Valley during the 2015 Gurkha earthquake.
4.2. Improvement of Infrastructure Resilience
A resilient infrastructure system is supposed to minimize the probability of failure, possess the redundant connectivity, shrink the recovery time, and limit impact propagation, which corresponds to four properties including robustness, redundancy, rapidity, and resourcefulness (the relationship between a resilient infrastructure and four properties of resilience is shown in
Figure 5) [
11]. A total of 45 papers investigated how the resilience of infrastructure systems can be improved and what strategies can be used to improve infrastructure resilience based on the four aspects. The majority of papers conducted the research on improving the comprehensive resilience of infrastructure resilience.
In terms of minimizing the probability of failure, the infrastructure systems are supposed to enhance the robustness and stability of responding to hazards or attacks. Several studies provided strategies for improving the resilience to resist various hazards using different methods [
55,
56]. Zhao et al. [
55] designed non-homogeneous hidden Markov models for resilience metrics in different damage scenarios such as natural disasters, man-made accidents, or violent attacks, aiming to propose an optimized scheduling strategy to improve and maximize the resilience of dynamic infrastructure systems. Johansen and Tien [
56] adopted the Bayesian Network methods for probabilistic modeling of interdependencies, which could probabilistically infer which interdependencies have the most critical impacts and prioritize components for repair or reinforcement, facilitating and providing strategies for improving the resilience of infrastructure systems when encountering natural attacks and targeted attacks.
There are previous studies dedicated to enhancing urban robustness using green infrastructure systems [
57,
58]. Staddon et al. [
57] stated that green infrastructure systems make great contributions to improving urban resilience; they conducted a high-level global overview of the contributions of green infrastructure to urban resilience and illustrated the challenges of reinforcing this green infrastructure-based pathway for seeking strategies of enhancing the urban robustness. Lee et al. [
58] simulated the impact of reduced stormwater runoff in the Gangnam district of Seoul through the application of green infrastructure and proposed a green infrastructure-based strategy on the basis of the flood-adaptive design strategy and simulation results for enhancing the resilience of urban infrastructure systems to the flood.
The interdependence and connectivity of systems are mutually beneficial under normal operating conditions, which also enhance the ability of urban infrastructure networks to resist extreme events, such as terrorist attacks and natural disasters [
59]. According to graph theory, connectivity is defined as the minimum number of elements (nodes or edges) that need to be removed in order to separate the remaining nodes into isolated subgraphs [
60]. Generally, a greater number of interconnecting paths between two nodes can contribute to the lower isolation and higher accessibility of the infrastructure system, as well as the greater redundancy of the infrastructure network [
61]. The performance of the infrastructure network can be improved as network connectivity grows, such as by increasing redundancy and adding the capacity of important links and interconnected nodes [
62].
As an extremely resilient system, it will recover rapidly after disruptions. Freeman and Hancock [
63] proposed distributed smart renewable energy micro-grid systems that mitigated adverse impacts by outage prevention and rapid service restoration, enhancing the resilience of energy and communication infrastructure in rural and regional Australia. Yang et al. [
64] proposed a multi-mode restoration model for enhancing the resilience of critical infrastructure systems to achieve the target of optimum post-disruption restoration under uncertainty, which included four steps: setting the boundaries of resilience analysis through prescribed standards; establishing the life cycle of infrastructure resilience management; defining physical-based infrastructure system functional modeling; designing an interface between interdependent infrastructure systems.
During the recovery period of disrupted infrastructure, rational resource allocation can minimize the economic and social impacts. Zhang, Kong, and Simonovic [
65] firstly proposed an optimal allocation model of infrastructure recovery resources, assisting decision-makers in understanding the effects of resources allocation better and to decide on the adoption of allocation strategies after a disruptive event. Kong, Zhang, and Simonovic [
66] then proposed a two-stage restoration resource allocation model to develop the optimal strategies for enhancing the resilience of interdependent infrastructure systems, with two goals about quickly restoring the dynamic resilience of infrastructure systems to meet the basic requirements of users in the first stage and minimizing the total loss in the subsequent recovery process in the second stage.