The use of Spearman’s rank correlation coefficients above 0.96 confirms that the new hybrid safety framework remains stable even during significant weighting shifts. This reliability is vital in the high-stakes environment of the oil and gas sector, where operational failures can lead to irreversible environmental damage and significant loss of life. As energy conglomerates in 2026 continue to outsource complex technical operations to third-party providers, the discrepancy between the safety cultures of various contractors has become a glaring vulnerability. Historically, the industry has relied on lagging indicators—metrics that record injuries or leaks only after they have occurred—to assess the competency of these external partners. However, such reactive methods fail to capture the subtle, underlying risks that precede a catastrophic event. This necessity for a more predictive, multidimensional oversight tool led to the development of the Contractor Health, Safety, Environment, and Quality Performance Index (HSE-Q PI), which integrates expert decision-making with machine learning to ensure that safety is managed as a proactive asset rather than a historical record. The result is a governance framework that allows for more granular oversight and targeted intervention across the global gas infrastructure landscape.
Theoretical Foundations: Safety as a Socio-Technical System
The architecture of this new performance index is deeply rooted in Safety Performance Theory and Socio-Technical Systems (STS) Theory, moving the conversation away from traditional administrative checkboxes. The researchers established that safety is an emergent property of a system, meaning it is the result of nonlinear interactions between technical components and human behavior. In the context of gas operations, this means that the reliability of a pipeline is not just about the quality of the steel or the pressure sensors, but also about the communication protocols and decision-making culture of the personnel managing those assets. By acknowledging these complex dynamics, the HSE-Q PI captures the reality of industrial environments where technical excellence can still be undermined by cultural or behavioral shortcomings. This shift in perspective is essential for organizations that manage vast networks of contractors with varying levels of maturity and internal governance, as it provides a holistic view of operational risk.
Furthermore, the integration of STS theory allows for a better understanding of how different organizational domains influence one another in real time. For example, a contractor’s commitment to internal auditing is not an isolated metric; it directly impacts their ability to identify emerging risks before they escalate into incidents. By utilizing a systems-thinking approach, the study moved beyond the siloed evaluation of metrics, instead focusing on how interdependencies drive safety outcomes. This theoretical transition justified the use of advanced mathematical modeling techniques that are capable of accounting for the feedback loops inherent in complex industrial settings. Instead of treating safety as a linear progression of tasks, the framework recognizes it as a dynamic state of balance that requires constant adjustment. This sophisticated understanding ensures that the resulting index remains relevant even as technical requirements and operational pressures evolve within the global energy market.
Defining the Indicators: A Standardized Safety Index
The initial phase of developing the HSE-Q Performance Index involved an extensive scoping exercise that synthesized international standards and two decades of specific industry data. Researchers meticulously reviewed guidelines from the Occupational Safety and Health Administration (OSHA) and the American Petroleum Institute (API) to identify 84 potential indicators that could reflect contractor performance. To ensure that these metrics were not only scientifically sound but also practically applicable in the field, a panel of fourteen experts with deep expertise in high-risk auditing was convened. This panel applied rigorous psychometric and content validity protocols to the candidate list, utilizing Lawshe’s methodology to filter out redundant or low-impact measures. The result of this refinement was a set of 43 validated sub-indicators that offer a comprehensive view of the health, safety, and environmental risks associated with gas operations, providing a level of detail previously missing from standard procurement checklists.
These validated sub-indicators were strategically organized into seven thematic dimensions to provide a structured approach to contractor evaluation. These dimensions include Awareness and Competence, Risk Assessment, Management Systems, Environmental Health, Monitoring, Emergency Preparedness, and Documentation. The internal consistency of this organizational structure was confirmed by a Cronbach’s alpha coefficient of 0.91, which signifies a very high degree of reliability across the entire index. This level of consistency means that the index can be applied across different geographic regions and operational contexts without losing its predictive power. By providing a standardized language for safety, the HSE-Q PI enables host organizations to compare contractors on an even playing field, ensuring that those with the most robust safety programs are recognized and rewarded. This standardization is a critical step toward raising the overall safety floor of the energy industry, as it forces all providers to meet a rigorous, data-driven set of expectations.
Implementing Precision: The Fuzzy Analytic Network Process
One of the most significant technical advancements in this framework is the application of the Fuzzy Analytic Network Process (FANP) to determine the weight of each safety indicator. Traditional evaluation methods often assume that different criteria are independent and that influence only flows in one direction, from the top of a hierarchy down to the specific tasks. However, real-world safety operations are characterized by a web of interacting factors where a change in one area, such as management commitment, can have a ripple effect across training quality and site inspections. The FANP model allows for the calculation of these intra-cluster and inter-cluster reciprocal influences, creating a more accurate representation of how safety actually functions on the ground. This mathematical sophistication ensures that the model does not oversimplify the complex nature of industrial risk, providing a more nuanced evaluation of which factors truly drive performance outcomes.
To address the inherent subjectivity that comes with expert opinions, the researchers utilized triangular fuzzy numbers within the FANP framework. Human judgment is often imprecise, with experts using phrases like “slightly more important” rather than specific numerical values. By assigning a range of values rather than a single fixed point, the fuzzy logic approach accounts for this ambiguity, ensuring that the final weights are more reflective of collective wisdom. These values were subsequently defuzzified to produce precise, actionable weights for each dimension of the index. The resulting consistency ratios averaged 0.054, which is well within the acceptable scientific threshold of 0.10. This confirms that the expert judgments used to build the model were logically consistent and that the framework is a reliable tool for administrative decision-making. By combining high-level mathematical modeling with real-world expert insight, the HSE-Q PI offers a bridge between theoretical risk assessment and practical operational management.
Prioritizing Drivers: Analyzing Weighting and Performance Gaps
The weighting results derived from the FANP analysis provided clear insight into what truly matters in high-consequence safety management. Awareness and Competence emerged as the most critical dimension, carrying 25.8% of the total weight, which underscores the belief that a well-trained, culturally aware workforce is the most effective defense against operational failure. Risk Assessment followed closely at 21.9%, emphasizing that proactive hazard identification remains more important than any administrative recording of past events. At the sub-indicator level, ISO certifications and formal safety training were identified as the strongest predictors of a contractor’s overall safety maturity. Interestingly, while documentation is often the focus of many regulatory audits, it was ranked as the least impactful direct driver of safety outcomes. This suggests that while paperwork is necessary for legal compliance, it should not be mistaken for the active, behavior-based safety measures that prevent accidents.
When the index was applied to real-world data from 200 active contractors, it revealed significant performance gaps that traditional audits often overlook. While many firms scored high in Environmental Health Management—averaging nearly 79 points out of 100—their performance in real-time Risk Management was notably lower, with a mean score of 61.16. This discrepancy indicates that while contractors are generally proficient at following environmental protocols and hygiene standards, they struggle more with the dynamic identification of operational hazards during active site work. This insight allows host organizations to shift their oversight strategies from general monitoring to targeted interventions. Instead of penalizing a contractor for a low overall score, a company can now identify specific weaknesses, such as a lack of hazard identification training, and work with the contractor to improve that specific domain. This collaborative approach fosters a more resilient supply chain and ensures that safety improvements are made where they are needed most.
Validation via Data: The Role of Machine Learning
To further validate the expert-driven findings, the research team employed a Random Forest regression model to analyze archival compliance data. This machine learning algorithm was used to determine which indicators most accurately predicted actual safety performance scores, providing an objective counterpoint to the subjective weights assigned by the experts. The model achieved a coefficient of determination of 0.62, which indicates a strong level of explanatory power and confirms that the features selected for the HSE-Q PI are indeed representative of real-world outcomes. One of the most compelling findings from this dual-layered approach was the alignment between machine learning and expert judgment regarding environmental management. Both methods identified Environmental Health as a critical factor, with weights of 22.5% and 21.0% respectively. This convergence adds a significant layer of credibility to the framework, proving that human intuition and algorithmic analysis can reach the same conclusions regarding industrial risk.
However, a fascinating divergence appeared regarding the importance of monitoring and auditing. While the expert panel assigned a relatively lower weight to auditing, the machine learning model found it to be a dominant predictor of short-term safety compliance. This difference can be explained by the varying time horizons of the two approaches. Experts often view training and awareness as the “cultural foundation” of safety—an essential investment whose benefits are realized over the long term. Conversely, data-driven models identify that frequent inspections and site audits have an immediate, measurable impact on reducing daily incidents. This realization led to the conclusion that an effective safety governance strategy must balance these two priorities. Organizations should invest in long-term cultural development through training while maintaining a rigorous, high-frequency auditing schedule to ensure daily adherence to safety protocols. By merging these two perspectives, the HSE-Q PI provides a balanced roadmap for both immediate compliance and sustainable safety excellence.
Actionable Next Steps: The Future of Safety Governance
The development and validation of the HSE-Q Performance Index established a new benchmark for how contractor safety is managed in the gas industry. The research concluded that a hybrid approach—combining the wisdom of human specialists with the objective predictive power of machine learning—provided a much more accurate risk profile than traditional lagging indicators. By moving beyond simple “pass or fail” pre-qualification checklists, organizations gained the ability to create detailed performance profiles for each partner in their supply chain. This allowed for a more strategic allocation of resources, where safety oversight was intensified for contractors in high-risk categories while those with proven maturity were given more operational autonomy. The sensitivity analysis further proved that the framework remained resilient under various conditions, making it a dependable tool for legal and administrative compliance in 2026 and beyond.
Future iterations of this governance model should focus on the implementation of dynamic weighting systems that adjust in real time as new data becomes available. As sensors and digital twins become more prevalent in gas operations, the performance index could be fed with live streaming data, allowing for even more proactive risk mitigation. This evolution would transition safety management from a periodic auditing process into a continuous, data-driven conversation between the host organization and its contractors. Furthermore, the methodology developed in this study is highly transferable to other high-consequence industries such as petrochemicals, mining, and large-scale construction. By adopting this hybrid analytical framework, industrial leaders can move toward a future where accidents are not just managed but systematically prevented through superior organizational design and technological integration. The lessons learned from this research provided the blueprint for a more integrated, transparent, and ultimately safer global energy economy.
