Fact-checked by Grok 2 weeks ago

Human reliability

Human reliability refers to the probability that a human performer will successfully complete a specified task without committing an error that could lead to system failure or unintended consequences.^[1] It is fundamentally the inverse of human error probability and is influenced by factors such as task complexity, environmental conditions, training, and performance shaping factors (PSFs) like stress or time pressure.^[2] Human reliability analysis (HRA) is a structured discipline that integrates systems engineering and behavioral science to identify, model, and quantify human contributions to overall system risk, particularly in probabilistic risk assessments (PRA).^[3] HRA aims to predict the likelihood of human failure events (HFEs) and recommend mitigation strategies, such as improved training, interface design, or procedural barriers, to enhance safety and reliability.^[2] By combining qualitative error taxonomies with quantitative probability estimates—often derived from operational data, laboratory studies, or expert judgment—HRA evaluates how human actions interact with hardware and software in socio-technical systems.^[3] The field originated in the 1950s at Sandia National Laboratories, initially applied to aircraft and nuclear weapons systems, with the first human reliability data bank established in 1962 by the American Institute for Research.^[3] Key developments in the 1970s and 1980s included the Technique for Human Error Rate Prediction (THERP) following the Three Mile Island accident, which emphasized PSFs, and subsequent advancements like the Standardized Plant Analysis Risk Human Reliability Analysis (SPAR-H) and Cognitive Reliability and Error Analysis Method (CREAM) in the 1990s and beyond.^[3] These evolutions reflect a shift from simplistic error probabilities to dynamic models accounting for context and cognition, as seen in post-Chernobyl analyses highlighting human factors in major incidents.^[2] HRA methods vary by approach: first-generation techniques like THERP and Human Error Assessment and Reduction Technique (HEART) focus on task decomposition and generic error rates, while second- and third-generation methods, such as A Technique for Human Event Analysis (ATHEANA) and MERMOS, incorporate contextual dependencies and recovery opportunities for more nuanced predictions.^[3] Data sources include incident reports, simulator exercises, and databases like the Nuclear Regulatory Commission's HRA database, enabling estimates where human error contributes to approximately 65-90% of incidents across high-hazard sectors such as nuclear, chemical, and aviation.^[2] Applications of human reliability span safety-critical industries, including nuclear power for licensing and design optimization, chemical and offshore oil for hazard mitigation, aviation for crew performance evaluation, healthcare for procedural error reduction, and manufacturing for operator-task reliability assessments.^[3]^[1] In railways and maritime sectors, HRA supports accident investigations and regulatory compliance.^[4] Emerging applications in cybersecurity and autonomous systems address human factors in cyber-physical interactions and vehicle reliability.^[5]^[6] Recent advancements as of 2025 include machine learning-enhanced predictive models and dynamic HRA for digital and advanced reactor environments, further integrating human performance with evolving technologies.^[7]^[8] Overall, HRA promotes resilient systems by prioritizing human-centered design and continuous performance improvement.^[9]

Fundamentals

Definition and Principles

Human reliability is defined as the probability that a human performs a specified task correctly within a given time under stated conditions, without committing errors that could lead to system failure.^[2] This concept is often quantified as a reliability measure R = 1 - P_e, where P_e represents the human error probability (HEP), emphasizing the probabilistic nature of human performance in complex systems.^[10] Key principles of human reliability distinguish between human error, which involves unintentional deviations from intended actions, and human failure, which may include intentional non-compliance such as violations of procedures.^[11] In socio-technical systems, where humans interact with technology and organizational elements, human reliability assesses the contributions of these interactions to overall system performance and safety.^[12] The HEP is fundamentally a function of performance influencing factors (PIFs) and task demands, such as environmental stressors or procedural complexity, which modulate the likelihood of errors.^[2] Human reliability plays a critical role in enhancing system reliability within safety-critical domains, where human errors contribute to 70-90% of failures in industries like nuclear power and aviation.^[13] For instance, analyses of aviation accidents indicate that 60-80% involve human error to some degree, underscoring the need to integrate human performance into risk assessments.^[14] The scope of human reliability encompasses both quantitative assessments, which estimate numerical probabilities of errors, and qualitative evaluations, which identify error modes and influencing conditions.^[12] It differs from human factors engineering, which primarily focuses on designing interfaces and environments to optimize human performance, whereas human reliability analysis predicts error likelihoods in operational contexts.^[15] PIFs serve as key modulators of reliability, with deeper exploration in subsequent sections on performance factors.

Historical Development

The roots of human reliability concepts emerged in the 1950s from advancements in behavioral sciences and reliability engineering, initially addressing human errors in military and aviation systems in the aftermath of World War II. Post-war analyses highlighted the limitations of treating humans solely as system components, drawing on engineering psychology to quantify error rates in man-machine interactions. This period laid the groundwork for systematic studies of human performance in high-stakes environments, shifting focus from mechanical failures to behavioral contributors. A pivotal milestone occurred in 1962 when Alan Swain presented the Technique for Human Error Rate Prediction (THERP) at a symposium of the Human Factors Society, formalizing methods for predicting human error probabilities through task decomposition and performance shaping factors. The 1970s saw accelerated development, spurred by the 1975 Reactor Safety Study (WASH-1400), which pioneered the integration of human reliability analysis (HRA) into probabilistic risk assessment (PRA) for nuclear facilities by estimating error contributions to accident sequences. The 1979 Three Mile Island accident further catalyzed this integration, revealing how human factors could exacerbate system failures and prompting regulatory endorsements for enhanced PRA methodologies. HRA evolved through distinct generations: first-generation techniques from the 1960s to 1980s emphasized static error probabilities derived from empirical task data; second-generation approaches in the 1990s incorporated cognitive psychology models to account for mental processes underlying errors; and third-generation methods from the 2000s emphasized dynamic contextual influences, performance recovery, and socio-technical systems. Major incidents like the 1986 Chernobyl accident exposed deficiencies in operator training and safety culture, driving advancements in the late 1980s and 1990s, while post-2000 developments, including responses to the 2011 Fukushima Daiichi accident, underscored human responses in extreme conditions. These events drove international standards, including IAEA guidelines for HRA in nuclear safety assessments that promote structured analysis of human and organizational factors. Concurrently, the field advanced toward data-driven HRA, leveraging empirical databases from operational events and simulations to refine error probability estimates beyond expert judgment. Recent developments as of 2025 include the integration of machine learning for predictive modeling of human error probabilities and specialized frameworks for advanced nuclear reactors with increased automation.^[7]^[16]

Performance Influencing Factors

Individual and Psychological Factors

Individual and psychological factors play a critical role in human reliability by influencing cognitive and behavioral processes that determine task performance and error likelihood. Psychological elements, such as stress and fatigue, can impair attention, decision-making, and response accuracy, while individual attributes like age, experience, and personality modulate susceptibility to these effects. These factors are intrinsic to the person and distinct from external influences, often compounding to elevate human error probabilities (HEPs) in high-stakes environments like nuclear operations or aviation.^[17] Stress, both acute and chronic, affects human reliability through its impact on arousal levels and cognitive function. The Yerkes-Dodson law posits an inverted U-shaped relationship between arousal (or stress) and performance, where moderate stress enhances efficiency for simple tasks, but excessive stress degrades performance on complex ones by overwhelming working memory and increasing impulsivity. In human reliability contexts, high stress can double or quintuple error rates, as seen in models adjusting HEPs for stressors like time pressure. For instance, acute stress from emergencies may lead to tunnel vision, reducing situational awareness, while chronic stress erodes long-term resilience to errors.^[18]^[19] Fatigue, influenced by circadian rhythms and sleep deprivation, similarly undermines reliability by diminishing vigilance and executive control. Circadian dips, particularly during night shifts, align with natural low points in alertness, exacerbating error propensity in prolonged operations. Sleep deprivation of 24 hours or more can impair psychomotor vigilance by causing lapses equivalent to blood alcohol concentrations of 0.10%, leading to slower reaction times and up to a threefold increase in errors on sustained attention tasks. Chronic partial sleep restriction compounds these effects, reducing overall cognitive throughput without full recovery during off-duty periods.^[20]^[21] Motivation and attitude further shape reliability, with low motivation fostering complacency that manifests as skill-based slips or violations. Complacency arises from overfamiliarity with routines, prompting reduced vigilance and normalization of deviations, as observed in maintenance tasks where operators skip checks due to perceived low risk. Positive attitudes, conversely, bolster adherence to procedures, but demotivating factors like role ambiguity can inflate error rates in repetitive scenarios.^[22]^[23] Among individual factors, age and experience exhibit a nonlinear influence on error rates. Novices exhibit peak error probabilities—often 5-10 times higher than experts—due to incomplete procedural knowledge and higher cognitive load during skill acquisition. Error rates decline sharply with 2-5 years of deliberate practice, stabilizing at low levels for mid-career professionals, but rise again in later years (post-60) from age-related declines in processing speed and memory. Physical health conditions, such as vision or hearing impairments, amplify these vulnerabilities; uncorrected visual deficits can elevate misperception errors in monitoring tasks by twofold, while hearing loss disrupts communication, contributing to coordination failures in team settings.^[24]^[22] Personality traits, particularly risk-taking propensity, also affect reliability by predisposing individuals to unsafe decisions. High risk-takers, often characterized by elevated extraversion and low conscientiousness, are more prone to rule-based errors in ambiguous situations, with studies in high-hazard industries showing 1.5-2 times higher violation rates compared to cautious peers. These traits interact with situational demands, where impulsive personalities under stress may bypass safeguards, heightening overall system risk.^[25] Quantification of these factors often involves HEP adjustments via performance shaping factors (PSFs) in human reliability analysis. For stress, multipliers range from 1 (nominal) to 2 (high) or 5 (extreme), applied to base error probabilities to reflect intensified cognitive strain. Fatigue under poor fitness for duty similarly yields multipliers of 2-3, while low experience can increase HEPs by factors of 3-10 depending on task type (action vs. diagnosis). These values derive from empirical data in probabilistic safety assessments, emphasizing scale without exhaustive benchmarks.^[26] Interactions among factors often compound unreliability, such as stress exacerbating fatigue in extended operations, where elevated cortisol amplifies sleep-deprived lapses, potentially multiplying error probabilities beyond additive effects. Organizational modulators, like training regimens, can mitigate these but remain secondary to intrinsic dynamics.^[21]

Organizational and Environmental Factors

Organizational factors play a critical role in shaping human performance reliability by influencing the broader context in which individuals operate. Safety culture, defined as the shared values and attitudes toward safety within an organization, significantly affects error reporting and learning from incidents. In a just culture, which emphasizes accountability for at-will choices while supporting error reporting without fear of unjust punishment, employees are more likely to disclose errors, enabling systemic improvements and reducing the recurrence of human failures.^[27] Conversely, a blame culture discourages open communication, leading to underreporting and persistent risks, as it attributes errors primarily to individual fault rather than systemic issues.^[28] Training and competence programs are foundational organizational elements that directly impact human error probability (HEP). Inadequate training or lack of experience can substantially elevate error rates, with multipliers in human reliability analysis (HRA) methods indicating increases of up to 17 times for unfamiliar tasks due to insufficient preparation.^[29] For instance, in the HEART method, error-producing conditions related to inexperience or lack of knowledge adjust baseline HEPs by factors ranging from 2 to 17, highlighting how organizational investment in ongoing competence development mitigates these risks.^[17] Similarly, workload and staffing levels influence reliability; understaffing often results in overload, where high workload as a performance influencing factor (PIF) can multiply HEPs by up to 5 in HRA assessments, as operators face divided attention and fatigue from excessive demands.^[29] Environmental factors encompass physical surroundings that can impair perceptual and cognitive processes, thereby degrading performance. Poor lighting conditions hinder visual tasks, increasing misreads and errors in information processing by contributing to PIFs that elevate HEPs through reduced signal detection.^[30] Extreme temperatures further compromise decision-making, with research showing that a 1°C rise in ambient heat can increase rational choice violations by approximately 1.1 percentage points, reflecting impaired cognitive function under thermal stress.^[31] System design, particularly human-machine interfaces (HMIs), also falls under environmental influences; poorly designed interfaces, such as ambiguous displays or non-intuitive controls, promote slips and lapses by failing to support error detection, often quantified in HRA as PIFs with multipliers up to 5 for interface inadequacy.^[29] Team dynamics within organizations mediate reliability through interpersonal interactions. Communication breakdowns, exacerbated by hierarchical structures, can prevent timely error correction; for example, authority gradients—where subordinates hesitate to challenge superiors—act as a PIF that suppresses feedback and heightens the risk of unaddressed errors.^[32] In HRA, such dynamics are modeled as organizational PIFs, adjusting HEPs to account for reduced team coordination. Quantification of these factors in HRA relies on PIF weighting schemes to modify baseline HEPs. Organizational inefficiency, including suboptimal safety culture or resource allocation, is incorporated as a composite PIF in methods like SPAR-H, where it can adjust probabilities by factors of 1.25 to 10 based on contextual severity.^[33] These schemes prioritize seminal approaches, such as those in HEART and CREAM, which integrate organizational and environmental elements to provide a holistic assessment of reliability influences.^[34]

Human Reliability Analysis Methods

First-Generation Techniques

First-generation techniques in human reliability analysis emerged in the 1960s and 1970s, primarily within probabilistic risk assessment (PRA) frameworks for the nuclear industry, focusing on decomposing tasks into basic elements to predict error probabilities using empirical data and expert judgment.^[35] These methods emphasized static, task-oriented models that quantified human error rates (HERs) through nominal probabilities adjusted by performance shaping factors (PSFs), such as stress or equipment design, but often neglected dynamic cognitive processes.^[29] The Technique for Human Error Rate Prediction (THERP), developed by Alan D. Swain in 1962 and detailed in the 1983 NUREG/CR-1278 handbook, represents a foundational approach.^[36] It involves step-by-step task analysis, breaking procedures into sequences of actions and verifications, such as reading instruments or manipulating controls, to identify potential error modes like omissions or commissions.^[36] Human error probabilities (HEPs) are calculated by assigning nominal rates from a database— for instance, 0.003 for reading an analog meter under good conditions—and multiplying by PSF multipliers (e.g., increasing error by a factor of 2-5 for high stress), while incorporating dependency modeling across tasks or team members via five levels from zero to complete dependence.^[36] Event trees integrate these to estimate overall task failure probabilities, with uncertainties captured using lognormal distributions and error factors (e.g., EF=10 for routine tasks).^[36] In the 1980s, the Human Cognitive Reliability (HCR) method, introduced by G.W. Hannaman and colleagues in an EPRI report, extended predictions to time-critical cognitive tasks by correlating error likelihood with response time ratios.^[29] It classifies operator behaviors into skill-based (automatic, low error), rule-based (procedural, moderate error), and knowledge-based (diagnostic, high error) categories, drawing from Rasmussen's SRK framework.^[29] HEPs are derived from time-reliability curves, where the probability of failure decreases as available time increases relative to a nominal response time— for example, a skill-based action completed in twice the nominal time might yield an HEP near 0.001.^[29] PSFs like stress or training adjust these curves, making HCR suitable for dynamic PRA scenarios in nuclear control rooms.^[29] Other PRA-based techniques include the Human Error Assessment and Reduction Technique (HEART), proposed by J.C. Williams in 1986, which simplifies analysis for broader applications.^[29] HEART categorizes tasks into 5-9 generic types (e.g., "totally unfamiliar task" with nominal HEP of 0.55 or "routine highly practiced task" at 0.0004) and identifies error modes such as omission or commission, then applies up to 38 error-producing conditions (EPCs) like poor interface design.^[29] The HEP is computed as the nominal rate multiplied by EPC multipliers (ranging from 1.3 to 50), weighted by the assessed proportion of affect (APOA, typically 0.5-1.0), providing a quick estimate without detailed task breakdown.^[29] Similarly, the Success Likelihood Index Method (SLIM), developed by D.E. Embrey and team in 1984, relies on structured expert judgment to scale PSFs.^[37] Experts rate 4-10 key PSFs (e.g., time urgency or procedures quality) on a 0-10 success likelihood scale, deriving a composite index that is logarithmically transformed into an HEP, often calibrated against benchmarks like 0.01 for moderately complex tasks.^[37] These techniques, while pioneering in quantifying HERs for PRA, share limitations as static models that overlook deeper cognitive contexts, dependencies beyond basic levels, and non-nuclear applications, leading to subjective adjustments and limited empirical validation outside controlled settings.^[29] Primarily tailored for nuclear risk assessments, they prioritize probabilistic prediction over behavioral simulation.^[29]

Second- and Third-Generation Techniques

Second-generation human reliability analysis (HRA) techniques, emerging in the 1990s, shifted focus from simple task-based error probabilities to cognitive processes underlying human performance, building on foundational probabilistic approaches by incorporating psychological models of error causation.^[35] A seminal framework in this category is the Generic Error Modeling System (GEMS), developed by James Reason in 1990, which classifies errors according to the stage of information processing where they occur. GEMS distinguishes between skill-based errors, such as slips (unintentional actions due to attentional failures) and lapses (memory failures), which arise during routine, automatic performance, and knowledge-based mistakes, which involve flawed plans or decisions in novel situations. This taxonomy emphasizes how cognitive demands and environmental cues influence error likelihood, enabling analysts to predict errors in dynamic settings rather than static tasks.^[38] Another key second-generation method is the Cognitive Reliability and Error Analysis Method (CREAM), introduced by Erik Hollnagel in 1998, which integrates cognitive engineering principles to assess error probabilities through performance modes and contextual factors. CREAM categorizes operator performance into four modes—strategic (proactive planning), tactical (rule-following), opportunistic (time-pressured actions), and scrambled (degraded control)—with error probabilities varying accordingly, such as approximately 0.00003 for strategic actions but up to 0.4 for scrambled ones under poor conditions. Central to CREAM is its Context Control Model, which evaluates nine performance influencing factors (PIFs), including working conditions, time available, and organizational support, each scored from -1 (adverse) to +1 (supportive) to adjust baseline error rates. This approach allows for a more nuanced quantification of human error probabilities (HEPs) by weighting cognitive and contextual elements, often reducing overestimation in complex scenarios compared to earlier methods.^[29] Third-generation techniques, developed from the late 1990s onward, further advanced HRA by embedding error analysis within probabilistic risk assessment (PRA) frameworks and emphasizing error-forcing contexts, particularly in high-stakes environments like nuclear power.^[35] The A Technique for Human Error Analysis (ATHEANA), sponsored by the U.S. Nuclear Regulatory Commission (NRC) in the mid-1990s, identifies unsafe acts by linking PRA-defined initiating events to potential human errors through triggers such as equipment failures or procedural ambiguities. ATHEANA then examines error-forcing contexts (EFCs), including time pressure and stress, to estimate HEPs, typically deriving them from empirical data rather than fixed tables, which supports tailored assessments in accident sequences. Similarly, the Method for the Representation of Errors under Stress (MERMOS), developed by Électricité de France (EDF) in the late 1990s for nuclear control room operations, focuses on stress-induced performance degradation by modeling operator cognition under emergency conditions.^[39] MERMOS uses a retrospective analysis of operator actions, classifying errors by cognitive functions like diagnosis and decision-making, and incorporates stress levels to adjust HEPs, drawing on simulator and incident data for validation.^[39] These generations introduced enhancements in modeling error recovery, recognizing that not all errors propagate unchecked, with probabilities of detection (P_d) often ranging from 0.1 to 0.9 based on monitoring opportunities and team dynamics, derived from empirical sources such as nuclear incident reports in databases like the Human Event Repository and Analysis (HERA).^[40] For instance, in CREAM and ATHEANA, recovery is factored into HEP calculations by assessing post-error detection via PIFs or EFCs, allowing for adjustments that reflect real-world mitigation. Data from nuclear events, including those compiled by the NRC, provide the empirical basis for these ranges, ensuring estimates align with observed recovery rates in control room simulations and actual incidents.^[40] Overall, second- and third-generation techniques offer advantages in handling dynamic, context-rich scenarios by accounting for cognitive depth and recovery, which reduces underestimation of HEPs in complex systems compared to first-generation task decompositions.^[35] This evolution enables more realistic risk assessments in industries with high human-system interaction, prioritizing influential cognitive and organizational factors over simplistic error rates.^[29] Emerging fourth-generation methods, developed since the 2010s and advancing as of 2025, incorporate dynamic modeling techniques such as Bayesian networks and machine learning to integrate real-time data and adaptive cognitive simulations, addressing evolving contexts in socio-technical systems like autonomous operations.^[41]

Specialized Frameworks

Specialized frameworks in human reliability provide structured approaches for classifying and investigating human errors after incidents, emphasizing the identification of causal chains rather than prospective probability estimation. These tools draw on performance influencing factors such as individual psychology and organizational conditions to dissect error pathways, facilitating targeted interventions in high-risk domains like aviation and nuclear operations. By focusing on retrospective analysis, they enable investigators to uncover latent failures that contribute to active errors, promoting systemic improvements over individual blame. The Human Factors Analysis and Classification System (HFACS), developed in the late 1990s by Scott A. Shappell and Douglas A. Wiegmann for U.S. military aviation, offers a hierarchical taxonomy for categorizing human contributions to accidents.^[42] Grounded in James Reason's Swiss Cheese model of accident causation, HFACS organizes failures into four tiers: organizational influences at the highest level, which encompass resource management, organizational climate, and process deficiencies; unsafe supervision, including inadequate oversight and failure to correct known problems; preconditions for unsafe acts, covering eight categories such as environmental factors (e.g., physical environment), operator conditions (e.g., fatigue), personnel factors (e.g., training), and team dynamics (e.g., crew resource management); and unsafe acts at the base, divided into errors (skill-based slips, decision mistakes, perceptual errors) and violations (routine or exceptional deviations).^[14] This structure, comprising 12 primary categories across the tiers, allows for comprehensive mapping of how latent organizational issues align with active frontline errors.^[43] Other notable frameworks include James Reason's classification of error modes, which distinguishes slips (observable execution failures, like pressing the wrong button), lapses (unobservable memory or attention failures, like forgetting a step), mistakes (flawed planning or problem-solving), and violations (intentional rule-breaking, either routine or exceptional).^[44] Originally outlined in Reason's 1990 seminal work Human Error, this typology serves as a foundational diagnostic tool for dissecting cognitive and behavioral breakdowns in incident reviews. Extensions of the Human Error Assessment and Reduction Technique (HEART), first proposed by J.C. Williams in 1985 for task-based error evaluation, have been adapted beyond probabilistic risk assessment (PRA) domains, such as in healthcare and construction, to qualitatively assess error-prone tasks by weighting influencing factors like experience and procedures without relying on quantitative probabilities.^[45] Similarly, the Information-Decomposition Analysis of Crew activities (IDAC) framework, developed by Ali Mosleh and colleagues in the mid-2000s, models crew performance through cognitive stages—information processing, decision-making, and action execution—in dynamic, team-based scenarios like nuclear control rooms, decomposing responses to stressors for post-event simulation and error tracing. These frameworks are primarily applied in post-incident investigations to trace latent failures back through causation chains, integrating performance influencing factors like stress or poor supervision as diagnostic inputs. For instance, HFACS analyses of aviation accidents have revealed that while unsafe acts appear in nearly 90% of cases, organizational influences underlie a substantial portion when examined hierarchically, contributing to broader systemic patterns. Unlike predictive human reliability analysis methods, specialized frameworks like HFACS and IDAC adopt a retrospective orientation, prioritizing the elucidation of error causation sequences over the quantification of failure probabilities. This diagnostic emphasis supports regulatory and organizational learning, focusing on holistic chains of influence rather than isolated event likelihoods.

Applications and Case Studies

Key Industries

In nuclear power, human reliability analysis (HRA) is integrated into probabilistic risk assessments (PRA) to evaluate operator performance during reactor operations, including shutdown sequences where errors in diagnosis and procedural execution can impact safety.^[36] For instance, HRA quantifies human error probabilities (HEPs) for such tasks, typically ranging from 0.01 to 0.1 under nominal conditions, adjusted by performance shaping factors like stress and training.^[36] This approach supports system reliability modeling and design improvements in control room environments.^[46] In aviation, crew resource management (CRM) training applies human reliability principles to mitigate crew errors by emphasizing communication, decision-making, and teamwork in high-stakes operations.^[47] CRM is employed in flight deck simulations to predict and reduce error rates through enhanced error trapping and recovery, contributing to a dramatic decline in accident rates since the 1980s.^[48] HRA techniques have also been applied in aviation simulations for specific scenarios, such as decompression events.^[49] The oil and gas sector, particularly offshore operations, focuses HRA on maintenance and drilling tasks where environmental performance influencing factors (PIFs) such as harsh weather increase error risks.^[50] Techniques like the Human Error Assessment and Reduction Technique (HEART) are adapted to quantify HEPs for these activities, incorporating error-producing conditions to prioritize interventions in dynamic settings.^[50] In healthcare, HRA methods predict errors in surgical teams by analyzing cognitive and organizational factors to enhance patient safety protocols.^[51] The Cognitive Reliability and Error Analysis Method (CREAM) has been adapted for this domain, identifying common error modes like misdiagnosis or procedural omissions that contribute significantly to adverse events, which affect approximately 10% of hospitalized patients globally (with about half deemed preventable).^[52] Similarly, in manufacturing, HRA supports error prediction along assembly lines by classifying failure modes in manual tasks, such as component misalignment, to improve process reliability and reduce defects.^[53] In railways, HRA is used for accident investigations and to assess operator reliability in signal passing and maintenance tasks, supporting regulatory compliance and safety enhancements.^[54] In the maritime sector, HRA evaluates crew performance in navigation and emergency response, incorporating factors like fatigue and bridge resource management to mitigate collision risks.^[55] Emerging applications include cybersecurity, where HRA models human errors in threat detection and response within IT operations centers as of 2023.^[56] In autonomous systems, such as unmanned aerial vehicles and self-driving cars, HRA addresses human oversight and handover errors, integrating with AI reliability assessments up to 2025.^[57] Cross-industry standards guide HRA implementation, with the International Atomic Energy Agency (IAEA) providing frameworks for integrating human factors into safety assessments across nuclear and analogous high-risk sectors.^[46] Complementary guidelines, such as those in ISO 31010 for risk management, recommend HRA techniques to address human factors in diverse industrial contexts.^[58] Methods like THERP serve as foundational tools in these applications for HEP estimation.^[36]

Major Incidents and Lessons

The 1979 accident at the Three Mile Island nuclear power plant in Pennsylvania exemplified operator misdiagnosis stemming from poor human-machine interfaces and high-stress conditions, where ambiguous instrumentation and inadequate training led to failure in recognizing a loss-of-coolant event. Human and organizational factors were identified as a root cause, with human errors contributing significantly to the accident's progression and partial core meltdown. Subsequent human reliability analysis (HRA) efforts, including retrospective applications of methods like MERMOS, underscored the need for enhanced operator training through realistic simulators to mitigate diagnostic errors under pressure. These insights prompted regulatory reforms, such as improved control room designs and emergency procedures, influencing nuclear safety standards worldwide. The 1986 Chernobyl disaster in Ukraine involved rule violations and knowledge-based mistakes by operators, exacerbated by organizational pressures that prioritized production over safety and fostered a deficient safety culture. Post-accident analysis using frameworks like the Human Factors Analysis and Classification System (HFACS) revealed how latent organizational failures, including inadequate training and suppression of safety concerns, enabled the escalation from a test procedure to a catastrophic explosion and release of radioactive material. This highlighted the role of performance-influencing factors (PIFs) such as poor communication and overconfidence in high-stakes environments. The findings drove international reforms, including strengthened safety culture assessments and redesigned reactor control systems to prevent unauthorized interventions. Beyond nuclear incidents, the 1977 Tenerife airport collision between two Boeing 747 aircraft demonstrated how communication errors, compounded by stress and hierarchical dynamics in the cockpit, can lead to runway incursions. Misunderstandings in radio transmissions, such as ambiguous phrases like "we are now at takeoff," occurred amid fog and time pressures, resulting in the deadliest aviation accident with 583 fatalities. Similarly, the 2010 Deepwater Horizon oil rig explosion in the Gulf of Mexico arose from maintenance oversights, including misinterpreted pressure tests and inadequate well control procedures due to confirmation bias and fatigue among the crew. These cases illustrate the pervasive impact of human factors across sectors, with lessons emphasizing enhanced error recovery protocols, such as standardized phraseology in aviation and rigorous well integrity checks in offshore operations. A key takeaway from these incidents is the value of mitigating PIFs through targeted interventions; for instance, the adoption of structured checklists in aviation has significantly reduced procedural errors by standardizing actions and minimizing omissions under stress. The broader impacts include a paradigm shift toward third-generation HRA methods following the 2011 Fukushima Daiichi accident, which exposed how seismic events impose cognitive stress, impairing situation awareness and decision-making amid infrastructure damage and radiation hazards. Empirical data from these major incidents have populated HRA databases, such as those compiled by the IAEA from event reports and simulator studies, enabling more context-sensitive error probability estimates and cross-industry applications for prevention.

References

[1]
[PDF] Human Reliability Assessment
Human Reliability Assessment. Human reliability is the opposite of human error. It is the probability of successfully performing a task.
[2]
[PDF] Human Reliability Analysis - Idaho National Laboratory
Definition of HRA. Human Reliability Analysis. • General Definition: A study of human contribution to overall risk when interacting with a system. − Part of ...
[3]
[PDF] Introduction to Human Reliability Analysis (HRA)
When combined with a system's reliability analyses, HRAs assess detrimental effects human errors have on the system. HRA involves error identification, modeling ...
[4]
A Narrative Review of Human Reliability Analysis (HRA) Techniques ...
This research comprehensively examines several HRA techniques, detailing their applications in a wide array of sectors, including railway, healthcare, ...Missing: sources | Show results with:sources
[5]
An Overview of Human Reliability Analysis Techniques ... - IntechOpen
Mar 13, 2013 · The objective of a human reliability analysis is 'to evaluate the operator's contribution to system reliability' and, more precisely, 'to predict human error ...<|separator|>
[6]
Human Reliability - an overview | ScienceDirect Topics
Human reliability analysis (HRA) aims at systematically identifying and analysing the causes, consequences and contributions of human failures in socio- ...
[7]
[PDF] Challenges in Human Reliability Analysis (HRA) - OSTI.GOV
Human reliability analysis (HRA) is used in the context of probabilistic risk assessment (PRA) to provide risk information regarding human performance to.
[8]
Managing human failures: Overview - HSE
Nov 4, 2024 · Managing human failures: Overview · Human failure aide-memoire ( PDF ) – This aide-memoire gives more information about the different failure ...
[9]
[PDF] How Many Performance Shaping Factors are Necessary for Human ...
The purpose of this paper is to explore the role of PSFs across different stages of HRA, including identification of potential human errors, modeling of these ...
[10]
human performance models and human errors in air traffic ...
Aug 9, 2025 · contributions (e.g. nuclear power - 70-90 ... human. functions. Error mechanisms and failure modes. depend on mental functions and knowledge which.
[11]
[PDF] Human Error and Commercial Aviation Accidents
Failed to correct known problems. The remaining two categories of unsafe supervision, the failure to correct known problems and supervisory violations, are ...Missing: nuclear | Show results with:nuclear
[12]
(PDF) Is Human Reliability Relevant to Human Factors?
Aug 10, 2025 · This paper presents a number of views from a panel discussion on the relationship between human reliability analysis (HRA) and human factors.
[13]
[PDF] Human Reliability Analysis Methods for Calculating Effects of ...
that, given the PIFs, there is a high probability of error. The quantification of the HEP was not considered to be a primary concern. Instead, the high HEP ...
[14]
https://www.faa.gov/sites/faa.gov/files/data_research/research/med_humanfacs/oamtechreports/200618.pdf
[15]
Cognitive, Endocrine and Mechanistic Perspectives on Non-Linear ...
The Yerkes-Dodson Law shows a non-linear relationship between arousal and performance, where high arousal impairs difficult tasks, but simple tasks are ...
[16]
Circadian Rhythms, Sleep Deprivation, and Human Performance
Both acute total sleep deprivation and chronic sleep restriction increase homeostatic sleep drive and degrade waking neurobehavioral functions.
[17]
Human reliability under sleep deprivation: Derivation of performance ...
One particular cause of fatigue is sleep deprivation; its effect on performance is not explicitly included in current HRA methods [23].<|control11|><|separator|>
[18]
[PDF] Human Factors - FAA Safety
Human factors, like fatigue and stress, contribute to aviation accidents, with 80% of maintenance errors involving them. Human error is a major cause of ...
[19]
The Human Factors "Dirty Dozen" | SKYbrary Aviation Safety
The Dirty Dozen refers to twelve of the most common human error preconditions, or conditions that can act as precursors, to accidents or incidents.Missing: reliability | Show results with:reliability
[20]
Human reliability and Organizational factors—How do Human ...
Results suggest that active errors, in the form of SPADs are influenced by factors as age, experience, average length of certified itineraries and previous ...
[21]
Mediating Effect of Risk Propensity between Personality Traits and ...
Feb 11, 2020 · This study takes a psychological approach and examines how personality traits affect unsafe behavior of construction workers.
[22]
[PDF] The SPAR-H Human Reliability Analysis Method
Based on review of first- and second-generation. HRA methods, the SPAR-H method assigns human activity to one of two general task categories: action or ...
[23]
Just Culture: A Foundation for Balanced Accountability and Patient ...
In a just culture, both the organization and its people are held accountable while focusing on risk, systems design, human behavior, and patient safety. Safety.
[24]
Fostering a just culture in healthcare organizations: experiences in ...
Aug 13, 2022 · A just culture is regarded as vital for learning from errors and fostering patient safety. Key to a just culture after incidents is a focus on ...
[25]
[PDF] Review of human reliability assessment methods RR679 - IChemE
Derbyshire. SK17 9JN. Human reliability assessment (HRA) involves the use of qualitative and quantitative methods to assess the human contribution to risk. ...Missing: authoritative | Show results with:authoritative<|control11|><|separator|>
[26]
[PDF] Guidance for Human Error Analysis (HEA)
Nov 21, 2019 · HEA enhances system reliability and safety by identifying where significant human errors could occur, the conditions that could provoke these ...
[27]
Hotter nights are disrupting sleep and hurting economic decisions
Jan 16, 2025 · Our findings revealed that a 1°C increase in midnight temperature was associated with a 1.1 percentage point rise in rational choice violations.
[28]
[PDF] Building a Psychological Foundation for Human Reliability Analysis
• Excessive Authority Gradient. Personality/Individual Differences PIFs. Examples. • Leadership style. • Deficiency in resource/task management. • Knowledge ...
[29]
[PDF] INL/EXT-10-18533, Rev. 2, "SPAR-H Step-by-Step Guidance."
These steps are: Step-1, Categorizing the HFE as Diagnosis and/or Action; Step-2,. Rate the Performance Shaping Factors; Step-3, Calculate PSF-Modified HEP; ...
[30]
Inter-relationships between performance shaping factors for human ...
Lastly, HEART [7] estimates final HEPs by multiplying nominal human unreliability with assessed effects which are calibrated from the multiplier of each EPC as ...
[31]
[PDF] Fifty Years of THERP and Human Reliability Analysis
Human reliability analysis (HRA) is the predictive study of human errors, typically in safety-critical domains like nuclear power generation (Swain and Guttman, ...
[32]
[PDF] NUREG/CR-1278, "Handbook of Human Reliability Analysis with ...
... human in a man-machine system that began after World War II and was influenced by the then new field of engineering psychology (Fitts, 1951a and b; Taylor ...
[33]
[PDF] NUREG/CR-4016, Vol. 1, "Application of Slim-Maud
The research conducted by BNL investigated one method of obtaining human reliability estimates from expert judges--the Success Likelihood Index Method-.<|control11|><|separator|>
[34]
Generic Error-Modelling System (GEMS) | SKYbrary Aviation Safety
EURCONTROL describes GEMS as "an error classification scheme developed by [Dr. James] Reason that focuses on cognitive factors in human error as opposed to ...Missing: paper | Show results with:paper
[35]
(PDF) MERMOS: EDF's new advanced HRA method - ResearchGate
Mar 1, 2017 · The research achievements in this paper can provide theoretical guidance to improve human error root reason analysis, an analysis basis of ...
[36]
[PDF] Human Event Repository and - Analysis (HERA) System
The objective of HERA is to make available empirical and experimental human performance data, from commercial nuclear power plants (NPPs) and other related ...
[37]
[PDF] A Human Error Analysis of Commercial Aviation Accidents Using the ...
The Human Factors Analysis and Classification System (HFACS) is a general human error framework originally developed and tested within the U.S. military as ...Missing: origin | Show results with:origin
[38]
[PDF] The Human Factors Analysis and Classification System--HFACS
Drawing upon Reason's (1990) concept of latent and active failures, HFACS describes four levels of failure: 1) Unsafe Acts, 2) Preconditions for Unsafe Acts, 3 ...Missing: origin | Show results with:origin
[39]
Human error: models and management - PMC - PubMed Central - NIH
They continually rehearse familiar scenarios of failure and strive hard to imagine novel ones. Instead of isolating failures, they generalise them. Instead of ...
[40]
Heart—A Proposed Method for Achieving High Reliability in Process ...
Aug 7, 2025 · It highlights those human errors, often stemming from inadequate training, poor communication, and flawed equipment design, significantly ...<|control11|><|separator|>
[41]
[PDF] Human Reliability Analysis for Nuclear Installations
The importance of HOFs in nuclear safety was highlighted by the Three Mile. Island accident in 1979, and HOFs were assessed to be a root cause of the accident.
[42]
[PDF] The Evolution of Crew Resource Management Training in ...
The research presented at this meeting identified the human error aspects of the majority of air crashes as failures of interpersonal communications, decision ...
[43]
[PDF] Successes and Failures in Civil Aviation - CORE Scholar
Crew Resource Management was widely adopted by U.S. airlines, and has generally been credited with helping to dramatically reduce their accident rates.
[44]
[PDF] Human Reliability Analysis for Oil and Gas Operations - arXiv
HEART was designed to be a quick and simple method for quantifying the risk of human error. – the method is designed to be used for HEP calculation only. HEART ...
[45]
[PDF] Human reliability analysis in healthcare: Application of the cognitive ...
Jun 4, 2012 · Cognitive Reliability and Error Analysis Method (CREAM) Since the stated goal of HRA is to improve reliability and safety, the ultimate goal is ...
[46]
Adverse Events - StatPearls - NCBI Bookshelf - NIH
Aug 2, 2025 · Adverse events are common across healthcare systems, with some studies' results reporting that 10% to 25% of patients are affected.
[47]
Classification and Quantification of Human Error in Manufacturing
The paper shows an application of human reliability analysis in a realistic manufacturing context to identify where and why manual assembly errors occur.
[48]
(PDF) Which Human Reliability Analysis Methods Are Most used in ...
Aug 31, 2023 · Human reliability Analysis (HRA) is a method to evaluate human error risk in critical safety tasks, according to influencing factors such as ...