Foundations¶
Operational Resilience Engineering (ORE) brings together vulnerability analysis, resilience engineering, operations research, and control theory to study how engineering systems can be understood and improved under disturbance, uncertainty, surprise, and changing operational demands.
ORE is organized around three linked but distinct analytical domains:
Vulnerability → Resilience → Control
This sequence is analytical rather than chronological. Vulnerability characterizes susceptibility to undesirable change and the perspectives used to evaluate it. Resilience characterizes the capacities, strategies, and designs through which acceptable performance can be preserved or improved. Control characterizes the processes through which systems sense, anticipate, act, and learn as conditions change.
The framework draws from established literature while making its own definitions, assumptions, and levels of analysis explicit. The goal is not to impose a single universal vocabulary for resilience research. It is to clarify which concept, measure, interpretation, and decision problem is being addressed in a given engineering analysis.
Vulnerability analysis¶
The first foundation of ORE is vulnerability analysis for engineering systems.
A vulnerability analysis begins with the question: under what conditions is a system susceptible to undesirable change? In engineering contexts, answering that question typically requires representations of scenarios or states, likelihood information, consequences, operational limits, feasible designs, and the relation between a system’s present and possible future performance.
Quantitative risk analysis has commonly used a triplet comprising scenarios, likelihoods, and consequences [KG81]. ORE builds on this tradition but distinguishes the underlying vulnerability data from the particular perspective used to evaluate it.
In the developing ORE formulation, a vulnerability space is represented by:
where \(S\) is a set of scenarios or system states, \(L\) contains likelihood information, and \(C\) contains consequence information. This space is not itself equivalent to risk. It becomes a quantitative decision framework only after likelihood and consequence information are interpreted and evaluated from a particular perspective.
This distinction matters because a system can appear vulnerable in different ways depending on whether the analysis prioritizes expected consequence, probability of failure, plausible worst-case disruption, or the possibility of crossing an unacceptable operational boundary. The current formal treatment of these perspectives is described on the Theory page and in Towards a Theory of Quantitative Vulnerability Analysis for Engineering Systems, available as arXiv:2610.02424 [EA26].
Risk, reliability, adversary, and safety¶
ORE distinguishes risk, reliability, adversary, and safety as related but non-interchangeable perspectives on engineering-system vulnerability.
Risk analysis commonly evaluates uncertain adverse consequences through a probabilistic interpretation of likelihood and a scalar interpretation of consequence. Reliability analysis commonly evaluates the probability that a system crosses a defined performance or failure boundary. Adversary analysis commonly emphasizes plausible worst-case consequences, including in interdiction and failure-budget models. Safety analysis emphasizes avoiding unsafe or unacceptable states and cannot generally be reduced to reliability or risk alone.
These distinctions are important because measures that appear numerically similar can preserve different information, prioritize different scenarios, and recommend different engineering designs. For example, a risk-oriented measure may prioritize reduction of rare catastrophic consequences, while a reliability-oriented measure may prioritize prevention of a more frequent crossing of an operational threshold.
ORE therefore treats risk as a perspective on vulnerability rather than as the umbrella concept that subsumes all engineering-system analysis. Reliability, adversary, and safety analyses are likewise distinct vulnerability perspectives. The current theoretical framework identifies their formal differences in likelihood and consequence interpretation.
This position is intended to address four recurring conceptual errors:
Conflation: treating risk, reliability, adversary, or safety as the same concept.
Subsumption: treating one perspective as generally a special case of another.
Substitution: using a measure from one perspective as though it measured another.
Reification: treating perspective-dependent measures as intrinsic properties of a system independent of an evaluative framework.
The framework does not imply that one perspective is universally correct. Instead, it holds that each perspective can reveal relevant vulnerabilities while also obscuring others. Consequently, a decision justified by one vulnerability perspective may impose tradeoffs relative to another.
Resilience engineering¶
The second foundation of ORE is resilience engineering. ORE uses resilience to mean a system’s ability to manage undesirable change, and resilient design to mean the engineering decisions intended to preserve or improve that ability.
This distinction connects resilience concepts to practical design questions: how is the system operating, what changes are possible, and what should those changes accomplish?
Resilience is not treated here as a direct synonym for risk, reliability, or vulnerability. Vulnerability identifies conditions under which undesirable change can occur and the perspectives through which those conditions are evaluated. Resilience concerns capacities, strategies, and designs that change how a system performs in relation to disturbance.
This distinction is particularly important for engineering decision-making. A vulnerability analysis can identify what conditions or scenarios deserve attention, but it does not by itself specify what intervention should be implemented. Resilience analysis evaluates how alternative designs, policies, configurations, or capabilities preserve, extend, recover, or improve acceptable performance.
A common engineering representation of resilience is a system-performance curve showing a selected measure of system function before, during, and after a disturbance. Such a curve can be useful as a phenomenological description of observed performance loss and recovery. However, it does not by itself explain why performance declines or recovers, what interventions were available or taken, or how alternative decisions could have changed the outcome [ESA25].
The limitation is not merely that a rebound curve is incomplete. A curve typically represents one selected performance measure and an observed trajectory after a disturbance; it does not necessarily represent system state, feedback structure, operational constraints, available control actions, adaptive capacity, or the decision process that produces performance over time. Consequently, the same observed curve can be consistent with materially different system dynamics and different opportunities for resilient intervention.
Recent work using linear time-invariant feedback control further demonstrates conditions under which a conventional system-performance curve falls short as a resilience model [DEP26]. ORE therefore treats system performance as an outcome of state stability, operational boundaries, design, and control processes—not as a curve that can by itself define resilience.
Woods’s four concepts¶
A central intellectual foundation is David D. Woods’s distinction among four concepts often grouped under the single label resilience:
Robustness: continuing to function in the presence of disturbance.
Rebound: recovering or returning toward prior performance after a disruption.
Graceful extensibility: extending adaptive capacity when surprise challenges the boundaries of ordinary operation.
Sustained adaptability: maintaining the capacity to adapt as demands, constraints, and conditions continue to change.
Woods’s work cautions against treating robustness and rebound as complete accounts of resilience, particularly when systems encounter surprise or exceed the limits of their ordinary adaptive capacity [Woo15, Woo18].
ORE agrees that robustness and rebound alone are incomplete. At the same time, it treats them as worthy engineering design strategies when the objective concerns system stability. Its developing contribution is to integrate state-stability strategies with boundary-oriented capabilities informed by graceful extensibility and sustained adaptability.
ORE builds on these distinctions by examining how resilience concepts guide engineering design. Its developing framework treats robustness, rebound, extensibility, and evolvability as distinct design perspectives, with explicit attention to system performance, state stability, operational boundaries, and available interventions.
The Theory page provides a conceptual introduction. The detailed resilient-design formulation is under development and will be released with a public preprint.
Surprise and infrastructure resilience¶
Critical infrastructure systems must operate amid changing demands, interdependencies, disturbances, constraints, and incomplete knowledge. Surprise is therefore not simply a forecasting error to be eliminated; it is a persistent operational condition for which systems, organizations, and operators must prepare.
Research on national and critical-infrastructure resilience has emphasized the integration of human and socio-technical capacities with technical-system processes [TESF19]. Later work further argues that infrastructure resilience must address surprise directly, including the ability to prepare for, respond to, and learn from conditions that exceed existing expectations [ADES22, SAP+25].
This literature motivates ORE’s emphasis on operational boundaries, adaptive capacity, feasible interventions, and the processes through which systems revise their understanding and performance following disruption.
Resilience terminology and scope¶
Resilience-related terminology varies substantially across infrastructure research and across disciplinary contexts. Reviews of critical-infrastructure literature show that terms such as resilience, robustness, reliability, vulnerability, adaptation, recovery, and sustainability are often defined differently or applied at different levels of analysis [MHS+23].
ORE responds by defining its concepts in relation to specific analytical objects:
Vulnerability concerns susceptibility to undesirable change and the perspectives used to evaluate it.
Resilience concerns capacities and design strategies that preserve, extend, recover, or improve acceptable performance.
Control concerns the processes that connect observation, interpretation, decision, action, and learning over time.
These definitions are intended to support comparison and integration where appropriate without assuming that superficially similar terms are equivalent.
Control and resilient performance¶
The third foundation of ORE is control: the continuing process through which a system observes conditions, interprets their significance, selects feasible actions, implements interventions, and updates its future behavior.
Control provides a formal language for distinguishing observed system performance from the mechanisms that generate it. The rebound curve is frequently used to summarize a system’s functional decline and recovery, but it is descriptive rather than explanatory: it does not identify why a system declined, why it recovered, what actions were taken, or what alternatives could have improved its performance [ESA25].
A performance trajectory may instead reflect the interaction of system state, disturbances, feedback, controller structure, operational constraints, resource availability, and intervention choices. Without representing these relationships, the same observed curve can be consistent with materially different system dynamics and different opportunities for resilient intervention. Feedback-control analysis demonstrates why a system-performance curve can omit the dynamics and control structure needed to explain, evaluate, or improve resilient performance [DEP26].
Within ORE, control is not limited to automatic feedback regulation. It also includes human, organizational, technical, and socio-technical processes that shape how an engineering system detects change, evaluates alternatives, acts under constraints, and learns from the resulting performance.
SAAL: sensing, anticipating, acting, and learning¶
ORE’s developing control formulation uses four connected processes:
Sensing: observing system performance, disturbances, resource conditions, constraints, and proximity to operational boundaries.
Anticipating: identifying possible future states, demands, threats, opportunities, and emerging boundary challenges.
Acting: selecting and implementing feasible interventions, operational strategies, policies, or designs.
Learning: revising models, interpretations, strategies, constraints, and future intervention choices in light of both successful and unsuccessful performance.
The SAAL processes provide an operational language for resilient control. They identify how information about system state and changing operational boundaries can enter a feedback process: sensing supplies observations; anticipating relates observations to possible future conditions; acting implements feasible control or design interventions; and learning revises the model, strategy space, and boundary assumptions used in future cycles.
The SAAL formulation is consistent with resilience-engineering accounts that connect resilient performance to the capacities to monitor, anticipate, respond, and learn. It also emphasizes the integration of human and socio-technical capacities with infrastructure-system processes [TESF19].
SAAL does not presume that surprise can be fully predicted, prevented, or eliminated. Instead, it organizes how systems and their operators can detect changing conditions, develop and revise expectations, implement feasible actions, and learn from the limits of prior models [ADES22, SAP+25].
Operations research and decision support¶
ORE uses operations research to formalize the decision structures that connect vulnerability, resilience, and control. These include:
Scenario and state spaces.
Likelihood and consequence information.
Performance measures and operational boundaries.
Constraints, resources, and feasible strategy spaces.
Engineering designs, policies, and interventions.
Tradeoffs among multiple vulnerability and resilience perspectives.
The purpose of formalization is not to suggest that all relevant system knowledge is fully observable, measurable, or reducible to a single metric. Rather, it makes assumptions and decision consequences explicit.
A central ORE proposition is that system data do not directly determine a decision. They must first be interpreted through one or more perspectives:
Data → Perspective → Decisions
This principle applies to vulnerability analysis, resilience evaluation, and control design. It also makes visible the potential for blind spots: a model can support effective action within its perspective while omitting vulnerabilities, capabilities, or consequences that would be salient under another perspective.
For resilient design, operations research also helps distinguish the assessment of current system performance from the selection of feasible interventions. An engineering decision model must make both its evaluative perspective and its intended mechanism of system change explicit.
Relationship among the foundations¶
The three domains are mutually connected:
Vulnerability analysis identifies how unacceptable performance can arise and which aspects of system behavior are visible under particular analytical perspectives.
Resilience analysis evaluates the capacities and design strategies that can preserve, extend, recover, or improve acceptable performance in relation to those vulnerabilities.
Control specifies the continuing sensing, anticipation, action, and learning processes through which vulnerabilities are recognized, resilience capacities are invoked, and models are revised over time.
ORE is under active development. The Theory page contains the current conceptual and formal architecture, while Research Path identifies planned technical notes, formal results, engineering examples, and research artifacts.
References¶
David L. Alderson, Rudolph P. Darken, Daniel A. Eisenberg, and Thomas P. Seager. Surprise is inevitable: how do we train and prepare to make our critical infrastructure more resilient? International Journal of Disaster Risk Reduction, 72:102800, 2022. doi:10.1016/j.ijdrr.2022.102800.
Tobias Demmer, Daniel Eisenberg, and Riccardo Patriarca. Why the system performance curve may fall short: a demonstration using linear time invariant feedback control. Environment Systems and Decisions, 46(4):78, 2026.
Daniel A. Eisenberg and David L. Alderson. Towards a theory of quantitative vulnerability analysis for engineering systems. 2026. URL: https://arxiv.org/abs/2610.02424, arXiv:2610.02424, doi:10.48550/arXiv.2610.02424.
Daniel A. Eisenberg, Thomas P. Seager, and David L. Alderson. The rebound curve is a poor model of resilience. PNAS Nexus, 4(3):pgaf052, 2025. doi:10.1093/pnasnexus/pgaf052.
Stanley Kaplan and B. John Garrick. On the quantitative definition of risk. Risk Analysis, 1(1):11–27, 1981. doi:10.1111/j.1539-6924.1981.tb01350.x.
Andrea Mentges, Lukas Halekotte, Moritz Schneider, Tobias Demmer, and Daniel Lichte. A resilience glossary shaped by context: reviewing resilience-related terms for critical infrastructures. International Journal of Disaster Risk Reduction, 96:103893, 2023. doi:10.1016/j.ijdrr.2023.103893.
Thomas P. Seager, Mazin H. AbdelMagid, Emily A. Pesicka, Daniel A. Eisenberg, and David L. Alderson. Infrastructure resilience to surprise. Risk Analysis, 45(12):5065–5079, 2025. doi:10.1111/risa.70161.
John E. Thomas, Daniel A. Eisenberg, Thomas P. Seager, and Erik Fisher. A resilience engineering approach to integrating human and socio-technical system capacities and processes for national infrastructure resilience. Journal of Homeland Security and Emergency Management, 16(2):20170019, 2019. doi:10.1515/jhsem-2017-0019.
David D. Woods. Four concepts for resilience and the implications for the future of resilience engineering. Reliability Engineering & System Safety, 141:5–9, 2015. doi:10.1016/j.ress.2015.03.018.
David D. Woods. The theory of graceful extensibility: basic rules that govern adaptive systems. Environment Systems and Decisions, 38(4):433–457, 2018. doi:10.1007/s10669-018-9708-3.