Contextual Evaluation of Risk Identification Techniques for Construction Projects: Comparative Insights and a Decision-Support Model

Abstract

Risk identification is a foundational process in construction project management, yet the selection of appropriate identification techniques often lacks empirical guidance. To address this gap, this study adopts a case study design and conducts a comparative evaluation of four established but underutilized methods—Delphi, Nominal Group Technique (NGT), Hazard and Operability Study (HAZOP), and Preliminary Hazard Analysis (PHA)—within the context of a large-scale infrastructure project in Türkiye. The Delphi panel consisted of five senior experts. The NGT session involved six site-level practitioners, and the HAZOP team was composed of four multidisciplinary professionals. Two project-level managers conducted the PHA. Each technique was assessed against seven evaluative criteria: methodological structure, stakeholder engagement, analytical depth, resource intensity, flexibility, decision-support value, and contextual fit. The findings reveal that HAZOP achieved the highest analytical depth and decision-support capacity, while NGT demonstrated the strongest stakeholder engagement and contextual adaptability. Delphi provided robust systemic insights but required substantial time and expert availability, whereas PHA offered rapid screening capacity with limited depth. Drawing on these findings, the study proposes a Contextual Decision Support Model that helps practitioners select the most suitable technique based on project complexity, available resources, and stakeholder conditions. This practical framework enables construction professionals to balance methodological rigor with contextual feasibility, ensuring that risk identification processes are both systematic and adaptable to real-world constraints. Beyond its methodological contribution, the study advances risk management in construction by providing a structured and transparent decision-support approach that bridges academic rigor with on-site practice. By aligning method selection with project-specific attributes and stakeholder dynamics, the model strengthens the integration of analytical precision and practical decision-making across the project lifecycle, thereby contributing to more proactive, evidence-based, and resilient risk management in construction projects.

1. Introduction

Construction projects are inherently exposed to multiple categories of risk. Financial risks—including cost overruns and cash-flow interruptions—have been widely observed in large infrastructure projects [1,2]. Safety and health risks, such as occupational accidents and hazardous exposures, remain persistent challenges [3,4]. Additionally, operational and project management risks, including delays, scope changes, and poor coordination, further exacerbate uncertainty [5]. Collectively, these risk categories underscore the crucial importance of systematic risk management in the construction industry. To mitigate such risks, systematic identification is essential, as it forms the foundation of any practical risk management framework.

According to the Construction Market Report [6] published by Research and Markets, the global construction industry is expected to reach a market value of approximately USD 17.04 trillion by 2025, driven by ongoing urbanization, infrastructure investments, and technological advancements in construction practices. This unprecedented size is primarily attributed to the increasing global demand for infrastructure development [7]. Nevertheless, large-scale infrastructure projects are frequently characterized by substantial technical complexity and financial volatility, rendering them particularly susceptible to diverse categories of risk. In this context, the strategic role of risk management has become increasingly pivotal in safeguarding project outcomes [8]. Absent a well-structured and systematically implemented risk management framework, fundamental project performance parameters—such as cost, time, and quality—remain vulnerable to disruption [9]. An overlooked or inaccurately assessed risk, for instance, may lead to significant cost overruns, ultimately transforming a potentially profitable investment into a financially unsustainable endeavor [10]. These multifaceted risks not only challenge individual projects but also shape the trajectory of the global construction sector, where unprecedented growth amplifies both opportunities and vulnerabilities.

Moreover, construction projects are inherently heterogeneous, distinguished by varying objectives, durations, contractual conditions, and financial constraints [11]. This intrinsic diversity implies that risk factors are neither uniform nor universally predictable across projects. Accordingly, risk management is increasingly acknowledged as an indispensable managerial function to ensure project alignment with predefined targets [12]. However, despite growing awareness of potential risks and their consequences among industry professionals, a substantial proportion of construction organizations continue to adopt fragmented or superficial risk management practices [13]. In many cases, risk management is perceived not as a value-adding strategic activity, but rather as a procedural burden that compromises time efficiency [14]. Against this backdrop, establishing a proactive, adaptive, and context-sensitive risk management system is essential for enabling successful project delivery under conditions of uncertainty.

Despite the increasing availability of risk management frameworks and international guidelines, their practical implementation within the construction industry—particularly in developing economies—remains inconsistent [15]. One of the primary barriers to institutionalizing risk management practices is the inadequate comprehension of the foundational stage: risk identification [16]. As the initial and arguably most critical phase of the risk management cycle, risk identification establishes the informational baseline upon which all subsequent analysis and decision-making depend [17]. Failure to accurately and comprehensively identify risks not only impairs the quality of risk assessment and response planning but also fosters a false sense of project control [10]. In this regard, the selection and application of appropriate risk identification techniques—tailored to the project’s complexity, scale, and stakeholder structure—are crucial.

Numerous techniques have been developed to facilitate systematic risk identification, ranging from conventional tools such as checklists and brainstorming sessions to more structured, expert-driven methodologies like the Delphi Method, NGT, HAZOP, and PHA [16]. While each technique offers distinct advantages, their practical utility varies significantly depending on project-specific factors such as organizational maturity, resource availability, stakeholder dynamics, and technical uncertainty [18]. Despite their potential, such methods are often underutilized due to a lack of methodological awareness, limited training opportunities, and cultural resistance to structured decision-making processes [19]. Prior studies consistently note that structured techniques remain underutilized in construction practice, mainly due to limited methodological awareness, lack of training, and cultural resistance to formalized procedures [4,5,20]. For instance, Hwang et al. [5] found that Singaporean contractors often perceived formal risk identification tools as time-consuming. Moreover, Zou et al. [4] highlighted similar patterns of low adoption in Chinese construction projects. Consequently, there is a pressing need to critically evaluate and compare the performance of these techniques across different project contexts to support evidence-based tool selection in risk-sensitive environments, such as construction. Based on the literature, three main research gaps can be identified:

• Few comparative implementations of Delphi, NGT, HAZOP, and PHA in construction projects, particularly in developing contexts.
• Limited investigation of contextual determinants, such as project type, institutional maturity, and stakeholder dynamics.
• Lack of operational selection tools that can guide practitioners in choosing the most appropriate risk identification technique.

These gaps underscore the need for a more thorough examination of existing risk identification techniques and their contextual applicability in construction projects.

Recognizing the pivotal role of systematic risk identification in safeguarding project outcomes—particularly in construction projects characterized by high uncertainty and institutional complexity—this study conducts a structured comparative assessment of four established techniques: Delphi, NGT, HAZOP, and PHA. Their empirical application within the context of a large-scale water treatment infrastructure project in Türkiye generates multiple analytical outputs, including benchmarking of methodological structures, strengths, and constraints; development of a contextual decision tree to support method selection; mapping of strengths and limitations against evaluative criteria; and visual representations such as radar charts and heatmaps to highlight relative performance and convergence patterns. Building on these results, the study pursues three objectives: (i) to evaluate the techniques across seven methodological and contextual criteria, (ii) to analyze the influence of project-specific factors on their applicability, and (iii) to develop a decision-support model to guide practitioners in method selection.

The remainder of this paper is organized as follows. Section 2 reviews the theoretical background and existing scholarship on construction risk management and risk identification techniques, outlining the key methodological families and emphasizing the gaps that this study seeks to address. Section 3 describes the research design and methodology, including the development of the multi-criteria evaluation framework, the systematic literature review process, and the case study protocol for applying the four structured techniques—Delphi, Nominal Group Technique (NGT), Hazard and Operability Study (HAZOP), and Preliminary Hazard Analysis (PHA). Section 4 presents the empirical results derived from the case study, summarizing the identified risks, their overlaps, and categorical distributions across contextual and occupational dimensions. Section 5 discusses the findings in depth, benchmarking them against regulatory data and prior studies, and comparatively evaluating the techniques along seven analytical criteria to reveal their contextual suitability and operational trade-offs. Building on these findings, Section 5.10 introduces the Contextual Decision-Support Model, which synthesizes comparative insights into a practical framework for selecting appropriate risk identification methods under varying project conditions. Section 6 summarizes the paper, discussing key implications, identifying limitations, and suggesting directions for future research. Finally, Section 7 elaborates the study’s contributions through three perspectives: theoretical (Section 7.1), methodological (Section 7.2), and practical (Section 7.3), highlighting how the research advances both academic understanding and professional practice.

2. Background

2.1. Risk Management in Construction Projects

A well-established risk management process typically consists of four sequential stages: risk identification, risk analysis, risk response planning, and risk monitoring and control [21]. Among these stages, risk identification serves as the cornerstone of the entire process, as it determines the breadth and quality of subsequent analyses and responses. Failure to identify risks adequately—either due to methodological limitations or organizational negligence—can compromise the effectiveness of the entire risk management framework, potentially leading to project delays, budget overruns, quality deficiencies, or reputational damage [22,23].

Despite the increasing global emphasis on formalized risk management frameworks, the level of adoption and institutionalization within the construction industry remains limited. Several empirical studies have highlighted that many construction firms either lack structured risk management protocols or apply them in a reactive and fragmented manner [24]. In many cases, risk management is perceived merely as a documentation or compliance requirement rather than a strategic management process. This perception is linked to the absence of in-house expertise, resource constraints, and the limited diffusion of risk-oriented organizational culture. Furthermore, the initial stage of risk identification is frequently overlooked or executed superficially, leading to incomplete risk registers that fail to reflect project-specific vulnerabilities [25]. Therefore, the selection and application of appropriate risk identification techniques remain a pivotal challenge in practice and a central concern of this study.

2.2. Classification of Risk Identification Techniques

To ensure comprehensive identification of potential threats and opportunities, various risk identification techniques have been developed in the literature, ranging from informal approaches to highly structured analytical methods. These techniques can be broadly classified into three main categories: conventional heuristic methods, group-based consensus techniques, and system-based analytical approaches [26,27].

• Conventional heuristic methods, such as checklists and brainstorming, are among the most frequently used tools in construction projects due to their ease of implementation and low resource requirements [27].
• The second category, group-based consensus techniques, includes structured expert judgment approaches such as the Delphi Method and the NGT. These methods aim to overcome the limitations of conventional tools by promoting systematic feedback, minimizing social bias, and enhancing group decision-making through iterative or structured procedures [28].
• The third category encompasses system-based analytical approaches, notably HAZOP and PHA. Originating from the chemical and process industries, these methods have been adapted for use in construction and infrastructure projects where technical complexity and safety concerns are prominent [29].

2.3. Risk Identification Methods

The Delphi Method is an iterative expert-elicitation process involving multiple rounds of anonymized questionnaires with controlled feedback [11]. In construction, it is beneficial where uncertainty is high and empirical data is scarce—such as in early-stage, large-scale, or innovative projects [30]. The method facilitates the identification of latent risks by allowing experts to refine their views after each feedback cycle. In practice, successful application requires well-structured questionnaires and the careful selection of domain-specific experts [31,32].

The NGT is a face-to-face, stepwise process that ensures equal participation through silent idea generation, round-robin sharing, clarification, and voting [33]. In construction projects involving multiple stakeholders—such as contractors, consultants, and clients—NGT allows rapid identification and prioritization of risks in a workshop setting [28]. It is most effective in small groups, where facilitator skill ensures balanced contributions and minimizes dominance bias [34].

HAZOP employs predefined guidewords to examine potential deviations from intended system functions [29]. In construction, it is primarily applied to technically complex projects (e.g., petrochemical facilities, tunneling, healthcare infrastructure) during the detailed design or pre-construction stages. It requires comprehensive design documentation and a multidisciplinary team to explore design and operational risks systematically [35]. Despite its analytical robustness, HAZOP is often underutilized in general construction practice due to its intensive resource requirements and the need for specialized technical knowledge.

PHA is a front-end screening method that identifies significant hazards early in the project lifecycle, using limited available data [10]. In construction, PHA is applied during conceptual design or feasibility phases to flag broad risk categories—such as safety, environmental, and operational hazards—before committing to detailed engineering [36]. It serves as a preparatory stage for more resource-intensive methods, such as HAZOP.

2.4. Research Gap

A broad spectrum of risk identification techniques is available to construction professionals, each characterized by distinct advantages contingent upon project complexity, resource availability, and stakeholder composition. Conventional methods, such as checklists and brainstorming, remain widely adopted due to their operational simplicity and ease of use. In contrast, more structured approaches—including the Delphi Method, NGT, HAZOP, and PHA—offer enhanced analytical rigor and systematicity. Despite the extensive documentation of the theoretical bases and procedural frameworks of these methods, comparative investigations exploring their practical applicability across various construction project typologies, particularly within developing country contexts, are still limited. A concise comparison of prior applications of Delphi, NGT, HAZOP, and PHA in construction projects is provided in

Table 1. Prior Studies on Risk Identification Methods in Construction.

Method	Key Studies	Context (Project/Domain)	Sample Characteristics	Outputs	Reported Limitations
Delphi	[16,30,31,32]	Early-stage, large-scale or innovative construction projects (high uncertainty, scarce empirical data)	Expert panels (domain-specific, iterative rounds)	Identification of latent/uncertain risks; refinement of expert judgments	Time-consuming; requires carefully designed questionnaires; dependent on expert availability
NGT	[33,34]	Multi-stakeholder construction projects (e.g., contractor–consultant–client workshops)	Small groups (6–10 participants, structured facilitation)	Rapid consensus on prioritized risks	Limited scalability; facilitator skill critical; potential bias if not well moderated
HAZOP	[29,35]	Technically complex projects (tunnels, petrochemical, healthcare infrastructure)	Multidisciplinary teams (designers, engineers, safety experts)	Systematic identification of deviations from intended design/operation	High resource and documentation demand; requires specialized training; often underutilized in general construction
PHA	[10,36]	Conceptual design or feasibility phases of construction projects	Small teams, often managers/engineers	Broad hazard screening across safety, environmental, and operational categories	Provides limited depth; mainly preparatory for advanced techniques (e.g., HAZOP)

to clarify the novelty of the present study.

Existing scholarship has predominantly emphasized single-method applications or has been restricted to narrow domains such as occupational safety or process engineering, often neglecting the multifaceted nature of real-world construction environments. The integrated deployment and comparative evaluation of these techniques within multidisciplinary risk identification processes thus represents an underexplored research frontier. Furthermore, prior studies seldom examine the influence of contextual determinants—such as project type, institutional maturity, and stakeholder dynamics—on the operational feasibility and effectiveness of each method. This study seeks to address these gaps through a structured comparative analysis of the Delphi Method, NGT, HAZOP, and PHA, situated within a large-scale infrastructure initiative: a water treatment facility in Türkiye. By grounding the evaluation in an actual project setting, the research not only synthesizes the methodological strengths, limitations, and contextual appropriateness of each technique but also generates empirically informed insights to guide method selection in complex and resource-constrained construction environments.

3. Materials and Methods

The methodology adopted in this study is structured around three interrelated components:

(1)

literature-driven development of evaluation criteria,

(2)

empirical data collection and synthesis through multiple structured techniques, and

(3)

comparative analysis and contextual model development (Figure 1).

Each component was carefully designed to ensure methodological rigor, transparency, and contextual relevance, particularly for construction environments characterized by heterogeneous project typologies and varying levels of institutional maturity in risk management.

Given the inherent complexity and uncertainty of construction projects—especially in emerging economies such as Türkiye—the study employed a multi-criteria analytical framework to evaluate the structural characteristics, operational requirements, and decision-support capacities of four widely applied risk identification techniques: Delphi, NGT, HAZOP, and PHA.

To establish a theoretical foundation, a comprehensive systematic literature review was first conducted to derive a set of evaluation criteria aligned with international risk management standards. These criteria captured seven analytical dimensions: methodological structure, stakeholder participation, analytical depth, resource intensity, flexibility and adaptability, decision-support capacity, and contextual fit.

Subsequently, empirical data were collected through the real-world application of the four techniques within a large-scale infrastructure project involving the design and construction of a drinking water treatment facility. Each method generated distinct sets of artifacts—Delphi produced R1/R2 survey responses and consensus statistics; NGT generated idea sheets, voting forms, and ranking tables; HAZOP yielded deviation logs and node–guideword worksheets; and PHA provided checklist-based hazard identification tables and risk matrices. All method-specific outputs were coded, consolidated, and integrated with the finalized evaluation criteria to form a unified comparative dataset, thereby ensuring a transparent data pathway from raw collection to criteria-based comparison.

In the final phase, these integrated data were comparatively analyzed across established criteria, leading to the identification of the strengths, limitations, and contextual suitability of each technique. The findings were illustrated using radar and heat-map visualizations and consolidated into a contextual decision-support framework that assists practitioners in determining the most suitable risk identification method based on project-specific factors such as uncertainty, technical complexity, resource constraints, and stakeholder interactions.

3.1. Research Design

Methodologically, the research is grounded in an interpretivist epistemological stance, which acknowledges the context-dependent, socially constructed, and experience-based nature of risk identification in construction projects [37,38]. Accordingly, the study does not aim to determine a universally superior technique but rather to understand the contextual appropriateness and operational alignment of each method within varying project environments, stakeholder compositions, and institutional capacities.

3.2. Literature-Driven Framework Development

This first component establishes the evaluation criteria that will later be integrated with empirical outputs from four structured techniques (Delphi, NGT, HAZOP, PHA) in Component 2 (see Figure 1).

3.2.1. Data Collection and Source Selection for Criteria Determination

A secondary data synthesis approach is employed, leveraging insights from peer-reviewed journal articles, books, industry reports, methodological handbooks, and empirical case studies published between 1995 and 2025. Following the PRISMA guidelines, the study selection process was carried out through a systematic, multi-stage procedure, ensuring the rigorous identification, screening, and inclusion of studies for the final synthesis.

In the first stage, keywords based on prior reviews, domain-specific handbooks, and authoritative frameworks, including ISO 31000 [20], the PMBOK Guide [21] (Project Management Body of Knowledge), and the APM Body of Knowledge (BoK) [39] were selected to ensure a strategic literature search process as the following: (“Delphi method” OR “Delphi technique” OR “Nominal Group Technique” OR NGT OR HAZOP OR “Hazard and Operability Study” OR “Preliminary Hazard Analysis” OR PHA) AND (“Risk Identification” OR “Hazard Identification” OR “Project Management”) AND (“Civil Engineering” OR “Construction Management” OR “Construction Process”). To ensure comprehensive coverage, the search was conducted in two types of sources: (i) peer-reviewed journal databases (Scopus, Web of Science Core Collection, ScienceDirect, and Google Scholar), and (ii) academic books and edited volumes dealing with construction risk management. This two-tiered approach ensured that the review encompassed both scholarly publications and foundational textbook knowledge on risk identification in the construction industry.

In the second stage, a set of inclusion and exclusion criteria was carefully established to guide the systematic screening and ensure methodological rigor. These criteria were designed to enhance transparency, minimize subjectivity in decision-making, and ensure that the final pool of studies is directly relevant to the review’s objectives. The inclusion and exclusion thresholds were determined based on established practices in systematic review protocols and tailored to the specific scope of this research. The exclusion and inclusion criteria are represented in

Table 2. Inclusion and Exclusion Criteria for Study Selection.

Inclusion Criteria	Exclusion Criteria
Studies offering methodological insights into the selected techniques. Applications of these techniques in real-world construction or infrastructure projects. Publications providing comparative evaluation dimensions—qualitative or quantitative.	Non-English articles without verified translations. Purely theoretical papers lacking empirical application. Studies confined to non-transferable industries outside the construction domain. Records retrieved from Web of Science or other databases that were not peer-reviewed publications (e.g., editorials, newsletters, non-scholarly sources).

.

In accordance with the PRISMA guidelines, a multi-stage procedure was followed to identify, screen, and select the studies included in this review. First, relevant articles and books were identified through systematic searches across multiple academic databases using predefined keywords. A total of 2565 records were initially retrieved from electronic databases. Second, duplicate records retrieved from different databases were removed, resulting in a comprehensive initial dataset, and 1136 records were captured. In the next stage, the titles and abstracts of all remaining studies were carefully examined to exclude those that were clearly irrelevant to the research scope. This step yielded a refined set of 38 candidate studies. Subsequently, the full texts of all articles in this candidate list were read in detail. Based on the pre-established eligibility criteria, 17 core references were ultimately selected to define the requirements. This structured approach ensured transparency, replicability, and alignment with the PRISMA framework, moving systematically from identification to screening, eligibility assessment, and final inclusion. The review process was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [40]. The detailed screening procedure is presented in Figure 2.

3.2.2. Development of Evaluation Framework

The construction and operationalization of the multi-criteria evaluation framework is presented, which is used to compare the selected risk identification techniques. To ensure methodological rigor and clarity, the framework is structured around three subcomponents: (i) criteria derivation and justification, (ii) alignment with international standards, and (iii) application strategy for comparative evaluation.

The evaluation criteria were synthesized from an extensive review of peer-reviewed literature, comparative methodological studies, and applied risk management research in construction and engineering contexts.

Methodological structure refers to the extent to which a technique is governed by a formalized process, structured steps, and predefined logic. A well-defined methodological framework enhances the repeatability, transparency, and reliability of risk identification outcomes. Chapman [27] emphasizes that structured approaches—such as facilitated workshops or guided checklists—provide greater consistency and reproducibility compared to ad hoc methods. Nawaz et al. [41] further highlight that the presence of a clear methodological structure reduces ambiguity and supports systematic decision-making in complex project environments. Similarly, Cooper et al. [42] argue that the adoption of formalized processes in risk management strengthens both the credibility and traceability of results, which is critical for organizational learning and stakeholder confidence.

Stakeholder participation assesses a technique’s ability to incorporate insights from diverse stakeholders, thereby enhancing the comprehensiveness, legitimacy, and acceptance of identified risks. Inclusive approaches are particularly valuable in collaborative and multidisciplinary construction environments, where risks often arise from the interaction of multiple actors and perspectives. Okoli and Pawlowski [31] emphasize that structured group techniques such as the Delphi method foster stakeholder engagement by enabling systematic input from experts across different domains. Hallowell and Gambatese [34] further demonstrate that broad participation not only enriches the pool of identified risks but also enhances the credibility of outcomes by reducing individual bias. Similarly, Albert et al. [43] argue that participatory strategies increase both the legitimacy and completeness of risk identification, especially in construction projects characterized by complex stakeholder dynamics.

Analytical depth refers to the extent to which a technique can capture detailed information, causal mechanisms, and interdependencies among risk factors. Methods with higher analytical depth are particularly valuable for identifying complex, latent, or systemic risks that may not be visible through superficial analysis. Tixier et al. [26] emphasize that advanced analytical methods are capable of mapping intricate interrelations among hazards, thereby providing more robust decision support. Similarly, Aven [22] argues that deep analytical approaches allow practitioners to move beyond probability assessments to understand causality and system-level interactions in uncertain environments. Furthermore, Hopkin [23] highlights that capturing hidden interdependencies is critical for large-scale projects, where risks often emerge from the interaction of multiple factors rather than isolated events. Consequently, high analytical depth is a defining attribute of techniques designed to address the complexity and uncertainty of modern project environments.

Resource Intensity denotes the level of human, financial, and temporal resources required for the practical application of a technique. Methods with lower resource demands are generally more feasible for small-scale projects or organizations with limited capacity, while highly resource-intensive approaches may only be viable for large, complex, or safety-critical projects. As Chapman [27] highlights, the selection of risk identification and assessment methods must account for practical resource constraints to ensure efficiency and realism in outcomes. Similarly, Tixier et al. [26] argue that different techniques vary significantly in their time and expertise requirements, which influence their suitability across various project contexts. Hallowell and Gambatese [34] further emphasize that structured, analytical methods—though often delivering deeper insights—tend to require greater resources in terms of expert involvement, data, and time. Therefore, aligning a technique’s resource intensity with the project’s size, scope, and organizational capacity is critical to maximize both efficiency and applicability.

Flexibility and adaptability refer to the technique’s capacity to adapt to variations in project size, phase, scope, and contextual conditions. Techniques that exhibit higher adaptability can be applied effectively across a broader spectrum of projects, thereby enhancing their practical utility. As noted in the PRAM Guide, the level of detail and application of risk management processes should be commensurate with project circumstances [39]. Similarly, the ATOM methodology explicitly differentiates practices for small, medium, and large projects, offering scalable activities aligned with project size [44]. Moreover, Valis [45] highlights that methods such as Delphi can be employed at various phases of the system life cycle, underscoring their flexibility across project stages. Contextual adaptability is further emphasized by De los Pinos and Garcia [46], who stress that single-method approaches may fail to capture project complexity; thus, methodologies must integrate diverse perspectives to remain effective in different project environments.

Decision support value refers to the extent to which a technique’s outputs facilitate actionable decision-making and can be seamlessly integrated into project planning and control processes. Techniques with a strong decision-support capacity not only generate risk registers but also provide structured insights for prioritization, response selection, and monitoring. As Hillson and Murray-Webster [47] emphasize, the ultimate purpose of risk management techniques lies in their ability to inform and improve managerial decision-making rather than merely documenting risks. Similarly, Aven [22] argues that methods with high decision-support value enable practitioners to translate complex risk analyses into practical strategies under uncertainty, thereby enhancing project resilience. Cooper et al. [42] also emphasize that risk analysis adds real value only when its outputs are systematically linked to planning, control, and performance measurement mechanisms, ensuring that decision-makers can effectively act on the identified risks.

Contextual fit assesses the degree to which a technique aligns with the institutional norms, cultural dynamics, and sector-specific constraints within a specific national or organizational context. A strong contextual fit ensures both local relevance and implementation feasibility, as risk management practices that ignore cultural or institutional realities often face resistance or become ineffective. Zou et al. [4] demonstrate that construction risk management frameworks must be tailored to reflect the regulatory, contractual, and cultural environments of the host country to achieve practical applicability. Similarly, Ling and Hoi [48] emphasize that cultural attitudes toward risk and collaboration directly influence the success of project risk management techniques in different national settings. Additionally, Smith et al. [49] argue that sectoral constraints and institutional practices define the adaptability and acceptability of techniques, making contextual alignment a decisive factor in their successful deployment.

Table 3. Evaluation Criteria for Comparative Framework.

Criterion	ISO 31000 [50]	Key References
C1. Methodological Structure	Refers to the degree to which a technique is governed by a formalized process, structured steps, and predefined logic. A robust methodological structure improves repeatability and transparency of risk identification efforts.	[27,41,42]
C2. Stakeholder Participation	Measures the technique’s capacity to incorporate inputs from diverse stakeholders, enhancing comprehensiveness and legitimacy of identified risks. Particularly important in collaborative and multidisciplinary construction environments.	[31,34,43]
C3. Analytical Depth	Assesses the level of detail, causality, and interdependencies captured by the technique. High analytical depth is crucial for uncovering complex, latent, or systemic risks in large-scale projects.	[22,23,26]
C4. Resource Intensity	Captures the demands imposed by the technique in terms of time, cost, expertise, and logistical support. Critical for method feasibility in resource-constrained environments.	[26,27,34]
C5. Flexibility & Adaptability	Refers to the technique’s ability to accommodate variations in project size, phase, scope, and context. More adaptable techniques are applicable across a broader range of projects.	[39,44,45,46]
C6. Decision Support Value	Indicates the extent to which the outputs of the technique facilitate actionable decision-making. A substantial decision-support value enhances the integration of risk outputs into project planning and control.	[22,42,47]
C7. Contextual Fit	Evaluates how well the technique aligns with institutional norms, cultural dynamics, and sector-specific constraints in local countries’ construction industry. Ensures local relevance and implementation feasibility.	[4,48,49]

presents the evaluation criteria used in the comparative framework, along with their operational definitions and key supporting references drawn from relevant literature in risk management, construction engineering, and systems analysis. The finalized criteria constitute the reference schema to which all method-specific artifacts will be coded and aligned in the empirical component.

To ensure content validity and global relevance, the criteria were mapped against two internationally recognized risk management standards:

• ISO 31000 [50]: 2018—Risk Management Principles and Guidelines
• PMBOK Guide (7th Edition) [21]—Project Risk Management Domain

This crosswalk ensures that the evaluation framework maintains both theoretical robustness and practical alignment with established best practices in risk management—for instance, C2. Stakeholder Participation aligns with Clause 5.4 of ISO 31000 [50] and the PMBOK’s [21] Stakeholder Engagement Principle, affirming its foundational role in participatory risk identification—similarly, C6. Decision Support Value corresponds to risk evaluation and prioritization functions detailed in Clause 6.5 of ISO 31000 [50] and the PMBOK [21] Decision-Making Environment.

Table 4. Crosswalk of Evaluation Criteria with International Risk Management Standards.

Criterion	ISO 31000 [50]	PMBOK Guide [21]	Explanation
C1. Methodological Structure	Clause 6.3—Risk Assessment Process	Tailoring Considerations	Emphasizes structured steps in method selection
C2. Stakeholder Participation	Clause 5.4—Communication & Consultation	Stakeholder Engagement Principle	Encourages participatory identification processes
C3. Analytical Depth	Clause 6.4—Risk Analysis	Risk Performance Domain	Depth of analysis in identifying risk sources and consequences
C4. Resource Intensity	Clause 4.3—Integration with Governance	Constraints and Resource Allocation	Considers the time, cost, and personnel demands of each technique
C5. Flexibility & Adaptability	Clause 4.2 & Principle 6—Continual Improvement	Tailoring, Complexity Dimensions	Methods’ ability to adjust to context or project scale
C6. Decision Support Value	Clause 6.5—Risk Evaluation	Decision-Making Environment and Data Models	Methods’ contribution to prioritization and actionability
C7. Contextual Fit	Clause 4.2—Understanding the Organization	Organizational Systems and Environment Fit	Compatibility with institutional, cultural, and regulatory context

aligns the selected evaluation criteria with internationally recognized risk management standards, specifically ISO 31000: 2018 [50] and the PMBOK Guide (7th Edition) [21], to ensure content validity and methodological consistency.

Each technique is qualitatively assessed against the criteria through a thematic coding and synthesis of empirical findings from the literature. Rather than assigning numerical scores, a narrative comparative approach is adopted to reflect the nuanced and context-dependent nature of risk identification in construction settings. This framework is also applied to a real-world case study, which serves as an empirical validation setting, allowing the evaluation framework to be tested under realistic constraints (e.g., time, stakeholder diversity, institutional readiness) and offering grounded insights on each technique’s feasibility and performance in practice. To further enhance robustness:

• Triangulation across multiple data sources and literature domains was performed.
• Operational definitions were developed for each criterion to ensure clarity and consistency in assessment.
• Reflexivity was maintained throughout the case study application to address subjectivity and contextual variability.

3.3. Empirical Data Collection and Case-Based Application

This second component operationalizes the study within a real-world infrastructure project and documents the full data collection streams for each technique. Specifically, Delphi generated R1/R2 survey responses and consensus statistics; NGT produced idea sheets, voting forms, and priority ranking tables; HAZOP yielded deviation logs, node–guideword worksheets, and risk register entries; and PHA provided checklist-based hazard identification tables, together with severity–likelihood matrices and ranking reports. All artifacts were subsequently subjected to data coding and synthesis (deduplication, semantic alignment of overlapping risks, and category mapping), after which the consolidated dataset was integrated with the finalized criteria (Figure 1). Templates and supporting materials for all four techniques are presented in Appendix A, Appendix B, Appendix C and Appendix D.

Developed a comparative evaluation framework that is grounded in a process-oriented, context-sensitive approach rather than a numerical ranking or scoring methodology. This decision reflects an epistemological commitment to interpretivism and acknowledges the inherent variability in construction project environments, stakeholder dynamics, and institutional capacities. Unlike prescriptive models that attempt to quantify performance hierarchies across techniques, the present framework emphasizes descriptive richness and operational nuance as core elements of methodological rigor.

Central to the evaluation strategy is the recognition that risk identification techniques do not operate in isolation from their implementation environments. Their effectiveness is shaped not only by their internal structures but also by external constraints such as resource availability, stakeholder diversity, project complexity, and regulatory maturity. As such, a rigid comparative metric would risk oversimplifying the socio-technical interdependencies that characterize real-world project settings.

Accordingly, the framework adopts a multi-criteria comparative lens, organized around evaluative dimensions (C1–C7), which were derived through a synthesis of academic literature and empirical practice (see Table 1). These dimensions—ranging from methodological structure to contextual fit—serve as interpretive categories rather than scoring axes. They guide a structured yet flexible assessment that accommodates the complex realities of infrastructure projects, particularly in emerging economies.

The evaluation was operationalized through a structured application of the selected techniques within an infrastructure project. However, the framework itself remains technique-neutral and adaptable, designed to surface the contextual compatibilities and trade-offs inherent in each method without privileging one over others. This strategy supports methodological pluralism while preserving analytical clarity, ensuring that the evaluation remains relevant across diverse project typologies and institutional contexts. This dual design approach ensured two critical outcomes: (1) methodological consistency across implementations, and (2) comparability of findings based on technique-specific characteristics rather than contextual discrepancies.

3.3.1. Participant Selection

To ensure credibility and expertise, participant recruitment followed strict inclusion criteria. Expert panelists were required to have at least ten years of professional experience in the construction sector, demonstrated familiarity with risk management practices, and had previously prepared or contributed to at least one formal risk management strategy in a construction project. For key roles (e.g., senior geotechnical engineers, project managers, crew leaders), a minimum of two participants with comparable responsibilities were consulted. This approach enhanced the diversity of perspectives and allowed for cross-validation of responses. Results were subsequently compared across participants occupying similar roles to identify consistencies and resolve potential discrepancies. To further reduce respondent bias, multiple participants were recruited for critical positions, and their ratings were systematically checked for internal consistency before consolidation.

Each method was then implemented with participants and documentation relevant to its operational requirements. For example, Delphi utilized iterative expert consultation via anonymized surveys; NGT involved a facilitated group workshop with frontline workers; HAZOP drew on technical schematics and guideword analysis in a multidisciplinary team setting; while PHA applied checklist-based screening during early-stage planning discussions.

Thematic coding of the evaluative criteria (C1–C7) was conducted by the author, with cross-checking performed by an independent colleague familiar with construction risk management. Disagreements in interpretation were resolved through discussion until consensus was achieved. Consistent with the interpretivist stance of the study, no numerical scoring was assigned; instead, the synthesis emphasized contextual themes, trade-offs, and situational appropriateness of each technique. Triangulation was employed by comparing three sources of evidence: (i) prior scholarly literature, (ii) project-specific documentation, and (iii) empirical observations from technique implementation. This multi-source comparison enhanced the robustness of the narrative synthesis, reducing single-source bias and increasing the credibility of the comparative evaluation.

3.3.2. Data Coding and Synthesis, and Integration with Evaluation Criteria

This stage served as the analytical bridge between the method-specific artifacts produced in the empirical component—namely, Delphi, NGT, HAZOP, and PHA—and the literature-driven evaluation framework, comprising seven criteria (C1–C7). The process unfolded as a sequential workflow involving data pre-processing, consolidation, thematic coding, integration, and quality assurance, ultimately resulting in a unified dataset ready for comparative evaluation and visualization.

All raw artifacts were first digitized, anonymized, and stored in a structured repository to ensure traceability. The corpus included Delphi R1/R2 survey responses and consensus statistics; NGT idea sheets, voting forms, and priority ranking tables; HAZOP deviation logs and node–guideword tables; and PHA checklist outputs, hazard identification tables, and risk matrices. These heterogeneous inputs were harmonized into a uniform tabular format, recording essential metadata such as method type, participant role, session date, risk statement, contextual tag, and evidence source. This initial harmonization established a consistent analytical foundation across techniques that differ significantly in documentation format and interaction structure.

In the consolidation stage, redundant and semantically overlapping items were merged to produce a canonicalized risk inventory. Duplicate detection was first performed using text normalization procedures (e.g., lemmatization and removal of repeated phrases). Then, conceptually equivalent risks described differently across techniques were semantically aligned under standardized labels while preserving a traceable back-link to all original entries. For instance, risks noted as “utility conflicts” in NGT and “unmarked utility strikes” in HAZOP were merged under a single canonical label once semantic congruence was verified. Ambiguous or context-dependent items were flagged and re-evaluated during subsequent coding rounds to prevent loss of contextual meaning.

A comprehensive codebook was then developed to guide thematic coding and ensure analytical consistency. The codebook included domain categories (e.g., technical, organizational, environmental), causal mechanisms (design error, logistics, regulatory constraint), and implementation attributes (e.g., facilitation intensity, documentation reliance). The first author conducted the initial round of coding, while a second researcher independently coded a stratified subset of the data to assess intercoder consistency. Any discrepancies were discussed and resolved through consensus, and the definitions in the codebook were refined accordingly. The finalized codebook was applied to the entire dataset, producing a multi-layered coded corpus that captured both the substantive and procedural dimensions of the risk identification outputs.

After thematic coding, all coded entries were integrated with the evaluation framework to align empirical evidence with the seven established criteria. For each technique, the coded data were examined to identify how its outputs reflected the attributes defined under C1–C7. Methodological structure (C1) was interpreted through indicators such as the degree of procedural formalization or documentation control evident in the artifacts, while stakeholder participation (C2) was inferred from the diversity and interaction intensity of contributors. Analytical depth (C3) was assessed through the richness and causality captured in outputs, and resource intensity (C4) was judged by the observable demands on expertise, time, and coordination. Flexibility and adaptability (C5) were derived from the ease with which each method could be tailored to varying project conditions, whereas decision-support value (C6) reflected the extent to which outputs translated into actionable planning inputs. Finally, contextual fit (C7) captured the alignment between technique outputs and the institutional, cultural, and regulatory setting of the case project. Rather than assigning numerical scores, this integration was carried out through qualitative comparative memos, each triangulating empirical observations, documentation evidence, and literature-based theoretical expectations.

To safeguard reliability, multiple quality assurance mechanisms were incorporated throughout the process. Each consolidated risk statement maintained a full provenance trail linking it to its original method-specific sources, ensuring auditability. Coding accuracy was reinforced through peer review of both the codebook and the thematic assignments. Moreover, member checking was implemented by returning synthesized outputs—such as Delphi consensus summaries, NGT priority lists, and HAZOP/PHA deviation logs—to participants for verification and confirmation. Reflexive memos were maintained to document interpretive decisions, and negative cases that contradicted initial assumptions were explicitly analyzed to preserve transparency.

The outcome of this process was a comprehensive, integrated dataset that merged qualitative and procedural evidence across all four techniques. This dataset consisted of a canonicalized risk inventory, a coded corpus with multi-dimensional tags, and a comparative evaluation matrix aligned with the seven criteria. These outputs provided the empirical foundation for the subsequent comparative analyses, radar and heat-map visualizations, and the synthesis of the contextual decision-support model that concludes the study.

3.3.3. Evaluation Procedure, Member Checking and Reflexivity

The evaluation of techniques against the seven criteria (C1–C7) was conducted through a structured interpretivist procedure. No numerical scoring was applied; instead, qualitative coding of expert panel responses and project documentation was performed. To mitigate researcher subjectivity, the first author undertook initial coding and subsequently reviewed it with an independent researcher who has expertise in construction risk management. Disagreements were discussed and resolved through consensus. To further reduce bias, cross-validation was applied across professional roles (e.g., project managers, safety specialists, design engineers), ensuring consistency of responses. Triangulation across expert inputs, project records, and prior literature provided an additional layer of verification. This stepwise procedure enhances both the transparency and replicability of the comparative evaluation.

To further strengthen methodological rigor, member checking was integrated into the iterative design of each technique. In the Delphi process, synthesized risk ratings from Round 1 were returned to participants in Round 2 as median and interquartile range (IQR) summaries, allowing experts to reconsider and adjust their judgments in light of collective feedback. Similarly, in the NGT workshop, aggregated prioritization results were presented to participants, who then engaged in clarification and re-ranking of the results. In the HAZOP and PHA sessions, consolidated deviation logs and checklist outputs were reviewed collaboratively to verify accuracy and completeness. These feedback loops enabled participants to validate synthesized outputs, thereby reinforcing the trustworthiness of the findings.

Reflexivity was also addressed explicitly. The author assumed a dual role, serving as both facilitator during the group sessions (Delphi questionnaire design, NGT moderation, PHA checklist review, and HAZOP node structuring) and as analyst during the synthesis of results. This dual involvement introduced potential sources of bias, such as influencing group dynamics or privileging specific interpretations during the coding process. To mitigate these risks, several safeguards were employed, including the use of standardized protocols (e.g., predefined guidewords, transparent rating scales, structured questionnaires), anonymized feedback structures, and cross-checking of coding outputs by an independent colleague with expertise in construction risk management. This reflexive acknowledgment underscores both the interpretivist orientation of the study and the safeguards implemented to minimize bias and ensure analytical transparency.

3.3.4. Project Background and Context

The selected case study for this research is a large-scale infrastructure project situated in the northwestern region of Türkiye. The project involves the planning and construction of a high-capacity water treatment facility. With a scheduled duration of approximately 24 months, the project encompasses various complex components, including civil works, mechanical and electrical installations, and the integration of advanced treatment technologies. Given its technical sophistication, regulatory sensitivity, and long-term strategic importance, the project represents a critical node in the region’s water management infrastructure. For confidentiality reasons, the specific identity of the project site is anonymized in this manuscript. Permission to report methodological applications was obtained from project stakeholders. Each risk identification technique was applied in alignment with its natural position within the project lifecycle—for example, the PHA was undertaken during the conceptual planning stage, Delphi rounds were conducted in the early design phase, NGT was organized during stakeholder coordination meetings, and HAZOP was implemented in the detailed design stage. This sequencing ensured contextual appropriateness without disclosing sensitive project timelines.

The project site is situated in a semi-urban area, characterized by a combination of residential development, environmentally sensitive zones, and logistical constraints resulting from limited access roads. Geotechnical investigations revealed variability in subsurface conditions, necessitating adaptive design considerations for excavation, foundation works, and pipeline routing. Furthermore, the project operates under the jurisdiction of multiple governmental agencies and involves collaboration among several private contractors, design consultants, and oversight institutions.

This multifaceted stakeholder environment introduces a variety of risk dimensions, including coordination complexity, regulatory uncertainty, and health and safety concerns during the construction phase. The nature of the project aligns well with the application of structured risk identification methodologies, as it involves both predictable risks typical of infrastructure projects and context-dependent risks arising from site-specific and institutional factors. The planning and execution phases required robust mechanisms to uncover, categorize, and prioritize hazards across different domains—technical, operational, environmental, and organizational.

3.3.5. Implementation of Risk Identification Techniques

The Delphi Method was applied through two iterative rounds involving a carefully selected panel of five domain experts, including two senior geotechnical engineers, one occupational safety consultant, and two project managers with over a decade of field experience. The first round aimed to elicit individual risk perceptions through anonymized questionnaires, while the second round facilitated convergence by presenting aggregated feedback for re-evaluation.

NGT was implemented through a structured 90 min workshop involving one site engineer, one safety supervisor, and four crew leaders from key disciplines (excavation, formwork, mechanical works, and logistics). Participants began by silently listing perceived risks, followed by a round-robin sharing process, clarification of listed items, and anonymous voting for prioritization.

HAZOP was implemented with a multidisciplinary team comprising a civil engineer, a geotechnical engineer, a safety coordinator, and a mechanical systems designer. The team analyzed the excavation and foundation phases using a simplified process flow diagram, employing key words such as “more,” “less,” “as well as,” and “reverse” to identify deviations from the design intent.

PHA was conducted through a semi-structured session between the project manager and the safety consultant, using a standard checklist adapted to the project’s lifecycle phases.

Table 5. Summary of the applied risk identification techniques.

Technique	Participants/Roles	Experience	Application Procedure	Session Duration	Artifacts Used
Delphi	2 Senior Geotechnical Engineers, 2 Project Managers, 1 Safety Consultant	>10 years each	Two iterative rounds; Round 1: anonymized questionnaires; Round 2: aggregated feedback for re-evaluation	Two rounds conducted over ~4 weeks	Structured R1/R2 questionnaires
NGT	1 Site Engineer, 1 Safety Supervisor, 4 Crew Leaders (Excavation, Formwork, Mechanical, Logistics)	10–12 years	90 min structured workshop; silent listing → round-robin sharing → clarification → anonymous voting	Full-day facilitated workshop	Idea sheets, voting forms, priority ranking
HAZOP	1 Civil Engineer, 1 Geotechnical Engineer, 1 Safety Coordinator, 1 Mechanical Systems Designer	10–15 years	Multidisciplinary team review; process flow diagram analyzed with guidewords (“more,” “less,” “reverse”)	One-day team session during detailed design	Guideword tables, deviation logs
PHA	1 Project Manager, 1 Safety Consultant	10–15 years	Semi-structured session using an adapted checklist across lifecycle phases	3 h session during the conceptual stage	Hazard checklists

summarizes the operational characteristics of each applied risk identification method, detailing session duration, participant composition, artifacts employed, and the number of risks identified.

4. Results

4.1. Quantitative Traceability of Risk Outputs

Although the present study is grounded in an interpretivist stance, quantitative traceability was also incorporated to strengthen methodological transparency. Across the four techniques, a total of 75 risks were identified: 17 through the Delphi method, 25 through the NGT method, 23 through the HAZOP method, and 17 through the PHA method. A cross-method comparison revealed that while several risks were shared, each technique also provided distinct insights (Table 5). For example, excavation collapse was consistently raised in NGT and PHA, while stakeholder conflicts were raised in NGT and HAZOP. At the same time, method-specific contributions were evident: Delphi uniquely highlighted seismic vulnerability and regulatory approval delays; NGT surfaced operational challenges such as logistics bottlenecks and supplier insolvency; HAZOP generated technically specific issues such as incorrect reinforcement placement and inadequate ventilation; and PHA emphasized early-stage risks, including delay in procurement of safety equipment and design errors in early drawings (Figure 3).

A data note was also included to clarify the derivation of these outputs: risks were first consolidated to remove duplicates, then mapped across methods based on semantic equivalence (e.g., “utility conflicts” in PHA and “unmarked utility strike” in NGT were treated as conceptually overlapping). Counts of unique and overlapping risks were subsequently generated, forming the basis of both the table and the Venn diagram. This combination of descriptive counts and visualizations ensures that the comparative evaluation retains its qualitative depth while providing a clear, quantitative trace of the evidence base. It should be noted that the evaluative criteria (C1–C7) were applied qualitatively, not numerically. This approach aligns with the interpretivist stance of the study and was supported by cross-validation, peer debriefing, and triangulation to ensure consistency.

4.2. Risk Identification

To provide a transparent overview of these patterns, the risks were organized into a comparative framework summarizing unique versus overlapping items. The results indicated that overlaps were most common in categories such as excavation hazards, stakeholder conflicts, and environmental exposures, whereas uniqueness was concentrated in areas reflecting the methodological strengths of each tool. These findings are presented both in tabular form (

Table 6. Summary of Method Applications in the Case Study.

Method	Total Risks Identified	Unique Risks	Overlaps with Other Methods
Delphi	17	15	2
HAZOP	23	20	3
NGT	25	20	5
PHA	17	13	4
Total (Overlapped Risks Excluded)	75

) and in a Venn approximation (Figure 3), which together illustrate the distribution of unique and overlapping risks across the four techniques. Axes, legends, and category labels were made explicit to ensure readability of the visuals.

4.3. Risk Categorization

In addition to mapping overlaps and unique contributions across methods, the identified risks were further categorized into two broad classes: occupational health and safety (OHS) risks (e.g., excavation collapse, electrical short circuits, hazardous material exposure, noise and dust exposure, worker fatigue, falls from height) and contextual or project management risks (e.g., budget underestimation, supplier insolvency, regulatory approval delays, stakeholder conflicts). Grouping the risks in this method clarifies their substantive domains while also illustrating how methodological design influences the breadth and emphasis of identified risks. This categorization is summarized in

Table 7. Categorization of the Identified Risks.

Method	Contextual/Project Management	Occupational Health and Safety
Delphi	Budget underestimation, Design–construction misalignment, Environmental opposition, High groundwater table, Insufficient site planning, Lack of early safety integration, Legal ambiguities in environmental permits, Potential contractor default, Regulatory approval delays, Schedule compression, Supply chain uncertainty, Unforeseen soil variability	Exposure to hazardous materials, Hazardous excavation conditions, Risk of deep excavation collapses in unstable soil, Seismic vulnerability of foundations, Slope instability in temporary works
HAZOP	Delayed material delivery affecting sequencing, Design error in mechanical piping system, Incorrect soil classification in design, Stakeholder conflicts	Collapse of excavation sidewalls, Compound failure modes, Dust exposure, Electrical short circuit during installation, Exposure to hazardous materials, Exposure to noise above permissible levels, Failure in dewatering system, Improper crane operation, Improper lifting gear selection, Inadequate ventilation in enclosed spaces, Incorrect reinforcement placement, Malfunction in groundwater dewatering systems, Mechanical equipment malfunction, Misalignment of pipeline supports, Overloading of retaining structures, Overloading of temporary scaffolding, Structural instability due to formwork failure, Unexpected groundwater inflow during excavation, Unmarked underground utility strike
NGT	Concrete delivery delay, Design change late stage, Logistics bottleneck, Material price escalation, Permit delay, Stakeholder conflicts, Supplier insolvency, Utility strike	Crane accident, Dust exposure, Electrical hazard, Equipment breakdown, Excavation collapse, Flooding on site, Formwork misalignment, Geotechnical instability, Mechanical system failure, Noise complaints, Rebar corrosion, Scaffolding failure, Welding defects, Worker fall from height, Worker fatigue
PHA	Caught-in or between, Delay in procurement of safety equipment, Design errors in early drawings, Environmental impact from excavation, Failure to obtain timely permits, Insufficient site planning, Stakeholder opposition during concept stage, Utility strike	Excavation collapse, Fall from height, Hazardous material mismanagement, Heavy equipment accidents, Inadequate emergency response planning, Material storage hazards, Noise and vibration disturbances, Undetected geotechnical variability, Struck-by

The items presented in italics denote risks that were identified through more than one method, thereby representing the overlapping set of risks across techniques.

, which highlights both the domain of each risk and the method(s) by which it was identified, thereby offering a transparent view of overlaps and complementarities across approaches.

5. Discussion

5.1. Validation Against Real-World Risk Data

A consistent picture emerges across regulatory and surveillance sources about the hazards that most frequently lead to severe outcomes in construction. In the United States, OSHA’s “Focus Four” (falls from elevation, struck-by, caught-in/between, and electrocutions) are identified as the leading causes of death on construction sites and remain the core emphasis of federal construction safety outreach and training [51]. Recent surveillance confirms this pattern: national injury-epidemiology syntheses from CPWR [52] show that fatal construction injuries continue to be dominated by falls from elevation, followed by struck-by and caught-in/between events, with electrical contact being a persistent contributor—together accounting for the majority of construction deaths in the last decade. The United Kingdom displays a similar distribution. In Great Britain in 2024/25, HSE [53] reported 35 fatal injuries in the construction sector, and among the “main kinds of fatal accident,” falls from a height accounted for 35 cases, followed by struck by moving object (18), trapped by collapse/overturning (17), struck by moving vehicle (14), and contact with machinery (13). Taken together, these converging data sources indicate that (i) work at height, (ii) impact/struck-by hazards (including plant, vehicles, and materials), (iii) caught-in/between or collapse events (e.g., trench/excavation, formwork), and (iv) electrical exposures represent the modal, high-severity risk channels in construction across jurisdictions. This external patterning provides a grounded benchmark against which to judge the realism of the risks elicited in this study’s techniques (e.g., repeated identification of excavation collapse, scaffolding/formwork failures, utility/electrical hazards, and dust/noise exposures), and it underpins the later mapping between method-specific findings and empirically dominant hazard classes in the sector.

Within the present study, many of these empirically dominant hazard categories were captured by the applied techniques. For example, excavation collapse and falls from height (related to concept scaffolding) were identified in both PHA and NGT, while scaffolding failure was identified in NGT and HAZOP. Electrical risks were consistently observed, with HAZOP surfacing “electrical short circuit during installation” and “exposure to noise above permissible levels,” while NGT emphasized “electrical hazard” and “welding defects.” Struck-by and collapse events were represented through risks such as “crane accident” (NGT), “structural instability due to formwork failure” (HAZOP), and “improper lifting gear selection” (HAZOP). These overlaps indicate a strong alignment between the model’s outputs and the most frequently fatal hazards identified by OSHA and HSE.

At the same time, the distribution of risk categories across methods reflects their methodological strengths. HAZOP’s structured guideword analysis, applied during the design stage, naturally generated detailed OHS hazards involving equipment malfunction, ventilation failures, and formwork instability, aligning with its emphasis on technical-system deviations. NGT, by contrast, emphasized operational risks directly observable on site—such as crane accidents, logistics bottlenecks, and worker fatigue—demonstrating its sensitivity to stakeholder engagement and site-level dynamics. Delphi primarily captured contextual and project management risks, including regulatory approval delays, supply chain uncertainty, and environmental opposition, reflecting its orientation toward eliciting expert judgment on systemic and organizational uncertainties. PHA, as expected from its screening function, surfaced early-phase risks such as inadequate emergency response planning, procurement delays, and concept-stage safety gaps, offering breadth rather than depth.

Beyond occupational safety, contextual and project management risks identified in the study also show strong resonance with established findings on project performance. International benchmarking studies [1,54] consistently report that supply chain uncertainty, design errors, and regulatory approval delays are leading contributors to cost overruns and schedule slippages. In this regard, the Delphi outputs—such as “budget underestimation,” “regulatory approval delays,” and “design–construction misalignment”—align with the high-frequency, high-impact drivers of project underperformance documented in the literature. This suggests that while Delphi may appear less focused on immediate OHS hazards, it adds critical value by surfacing systemic risks with significant downstream severity for project delivery.

Consequently, the study’s elicited risks are not only internally coherent but also externally validated against high-frequency, high-severity categories consistently observed in regulatory statistics and project risk research. While the alignment here is interpretive rather than statistical, it nonetheless provides credible evidence that the proposed model captures the empirically dominant hazard channels in construction. The triangulation between elicited risks, OSHA/HSE fatality statistics, and project risk literature enhances confidence in the contextual realism of the model.

5.2. Comparative Evaluation of Structured Risk Identification Techniques

The comparative evaluation of the Delphi Method, NGT, HAZOP, and PHA within this study yields several findings that both corroborate and refine existing knowledge in the domain of construction risk identification. For instance, previous research by Kiral et al. [16] highlighted the methodological robustness of the Delphi Method in identifying context-specific risk factors in Turkish construction projects. The study’s findings align with this assertion, particularly in terms of the technique’s analytical depth and its ability to accommodate expert tacit knowledge. However, unlike Kiral et al. [16], this study emphasizes the technique’s limitations in terms of implementation time and resource intensity, especially under conditions of rapid decision-making or constrained project timelines. In the case application, while Delphi enabled the identification of latent, cross-disciplinary risks—such as delayed permit approvals and supply chain delays—the method’s multi-round structure led to participant fatigue, with later rounds yielding diminishing engagement and reduced response richness. Moreover, the technique excelled in identifying systemic and latent risks—such as legal ambiguities in environmental permits, risk of deep excavation collapses in geotechnically unstable soil, and potential contractor default. However, challenges emerged in sustaining participant engagement across multiple rounds, coordinating schedules, and ensuring consistent interpretation of risk semantics. Furthermore, the need for a skilled facilitator and sufficient time allocation posed constraints, especially under projects with fast-tracked timelines.

Similarly, while Hallowell and Gambatese endorsed the NGT as an efficient method for safety-related hazard identification, particularly for its inclusivity and participatory structure, the present study reveals that its decision-support capabilities may be diminished in settings characterized by stakeholder asymmetry or hierarchical imbalance [34]. Unlike brainstorming, which was critiqued by Tixier et al. [26] for its lack of structure, the NGT offers procedural rigor; yet, the current findings suggest that the absence of anonymity may introduce social desirability bias or deference to authority figures within construction teams—especially in culturally high power-distance contexts. During the on-site NGT workshop, for example, junior crew leaders appeared reluctant to challenge or override senior staff input, despite divergent perceptions of risk priorities. While the technique facilitated shared ownership of risk perception, it struggled to capture dissenting views under hierarchical pressure. Its inclusive and participatory structure encouraged open communication and fostered a sense of psychological ownership among workers. However, the focus on immediate and observable risks tended to overlook upstream or strategic threats, and the lack of complete anonymity may have discouraged junior participants from expressing dissenting views in the presence of supervisory personnel.

Regarding system-based methods, prior studies (e.g., [35,55]) have generally positioned HAZOP as a best-practice standard for high-risk technical environments. While the present study concurs with these conclusions concerning analytical rigor and depth, it also highlights that HAZOP’s extensive resource demands and reliance on advanced design documentation render it less suitable during early-stage project planning. In field implementation, HAZOP yielded detailed technical failure modes, but required significant facilitation expertise and exhaustive pre-meeting preparation, underscoring its practicality only in mature project phases with substantial documentation in place. The technique demonstrated impressive analytical depth and procedural rigor, particularly in surfacing failure scenarios that could trigger cascading effects. However, its implementation required detailed design documentation, significant preparation time, and technical fluency among team members—conditions that may not always be available in early-stage or resource-constrained projects. Additionally, the abstract nature of guidewords occasionally confused less experienced participants, requiring frequent clarification by the facilitator.

In contrast, PHA is affirmed as a lightweight and agile method for preliminary assessments, echoing earlier insights by Merna and Al-Thani [10]; however, this study further reveals its vulnerability to generalization and its limited integration with downstream risk control mechanisms. The PHA session, conducted during the project’s initial planning phase, identified high-level concerns but lacked granularity for actionable task-level mitigation, requiring further elaboration. The strength of PHA lies in its speed, simplicity, and minimal resource requirements, making it particularly useful during early project planning or tendering stages. However, the method lacked depth in risk interaction analysis, failed to incorporate stakeholder diversity, and provided limited utility for downstream mitigation design. Its tendency to generalize and oversimplify complex risk scenarios also reduced its predictive power in the face of system-level failures.

5.3. Comparative Evaluation Matrix

In terms of methodological structure, the Delphi Method offers a highly organized and iterative framework, in which expert opinions are refined across multiple rounds through anonymized feedback [31,56]. This design enhances the reliability of consensus and is particularly useful in situations marked by high uncertainty and limited data. NGT also follows a structured approach but emphasizes a more condensed, face-to-face format involving silent idea generation, round-robin sharing, and ranked voting [34,57]. HAZOP presents the most rigid methodology among the four, relying on systematic guideword analysis of design components [55]. It is process-intensive but ensures exhaustive hazard identification. In contrast, PHA employs a loosely structured, qualitative approach that supports rapid identification of broad hazard categories without procedural complexity [10].

Regarding stakeholder participation, Delphi allows for widespread and geographically dispersed expert engagement due to its asynchronous and anonymous nature, making it ideal for national or international studies [31]. NGT, by contrast, is most effective in small, physically co-located groups, where balanced interaction can be facilitated in real-time [34,57]. HAZOP requires a multidisciplinary team with technical expertise to analyze system components in live sessions [55] collaboratively. PHA typically involves fewer stakeholders, usually technical staff or project managers, and is conducted early in the project lifecycle with minimal coordination requirements [10].

When evaluated on analytical depth, HAZOP emerges as the most robust technique due to its systematic node-by-node inspection of potential deviations from design intent [35,55]. Delphi also provides considerable depth by aggregating nuanced expert insights, although it leans more on subjective judgment than technical breakdown [31]. NGT delivers moderate analytical depth by ranking group-generated risks but lacks iterative refinement [57]. PHA ranks lowest in this category, as its scope is broad and often limited to qualitative classification of significant hazards [58].

In terms of resource intensity, HAZOP demands the highest level of investment, requiring detailed documentation, skilled facilitation, and time-intensive workshops [35,55]. Delphi is also resource-heavy, mainly due to the need for participant recruitment, round design, and multiple feedback cycles [31]. NGT, on the other hand, is relatively economical, requiring only a trained facilitator and a small participant group to execute effectively [34,57]. PHA is the least resource-intensive, enabling rapid deployment with minimal personnel and documentation [15].

With respect to flexibility, Delphi is highly adaptable, suitable for a wide variety of project types and participant configurations [31]. NGT is moderately flexible, though it is constrained by group size and the need for physical presence [57]. HAZOP is the least flexible due to its dependence on detailed system design and formal guideword protocols [55]. PHA is the most agile technique, as it can be conducted with limited project data and easily adapted to different project phases and team compositions [10].

In terms of decision-support capability, HAZOP provides the most comprehensive output, particularly for design optimization and safety assurance. Its findings are often directly linked to engineering actions [55]. Delphi also supports decision-making effectively by consolidating expert knowledge into prioritized risk lists [25]. NGT facilitates group consensus on key risks but lacks depth in technical recommendations [28]. PHA provides high-level guidance during the conceptual phase but is less effective at supporting detailed decisions due to its generalist nature [35,58].

Finally, in terms of contextual fit, PHA and NGT stand out as practical tools due to their simplicity, low resource demands, and suitability for environments where formal risk management practices are still in the process of evolving [10,34]. Delphi, while methodologically robust, tends to be more feasible for academic or policy-level initiatives given its iterative structure and coordination intensity [31]. HAZOP, despite offering the highest degree of technical rigor, is often underutilized in settings with limited institutional capacity or documentation infrastructure [55]. Its application is typically reserved for high-risk, technically complex projects, such as energy facilities or large-scale industrial systems [35,59]. The comparative performance of the four techniques across the selected criteria is summarized in

Table 8. Characteristic of the Methods.

Criterion	Delphi Method	NGT	HAZOP	PHA
C1. Methodological Structure	High—Iterative, formal, structured questionnaire	Medium—Structured steps, less iterative	Very High—Guideword-based, standardized process	Medium—Checklist-style structure
C2. Stakeholder Participation	Moderate—Limited to selected experts	High—Interactive, inclusive of all participants	Moderate—Technical team focused	Moderate—Early-stage multidisciplinary teams
C3. Analytical Depth	High—Aggregates expert judgment and refinement	Medium—Focused on idea generation and ranking	Very High—Identifies design deviations and hazards	Low to Medium—Identifies broad categories only
C4. Resource Intensity	High—Time-consuming, requires multiple rounds	Low—Quick sessions, minimal prep	Very High—Requires technical documents and facilitation	Low—Fast, minimal documentation required
C5. Flexibility & Adaptability	High—Can be customized across domains	High—Easily modified based on scope	Low to Medium—Suited for detailed designs only	High—Flexible for early-phase application
C6. Decision Support Value	High—Informed by expert convergence	Medium—Group consensus but less analytical	Very High—Detailed outputs for design-level decisions	Medium—Used for early-stage screening
C7. Contextual Fit	Medium—Requires expert availability	High—Works well in workshops or training groups	Low—Demands high technical maturity	High—Suitable for low-resource, early-stage use

, providing a side-by-side evaluation to guide method selection across a wide range of construction project environments.

To strengthen the comparative dimension of the analysis, the distinctive characteristics of the four techniques were synthesized across the seven evaluation criteria (C1–C7). The Delphi Method and NGT both emphasize structured group-based elicitation, yet they differ substantially in terms of process dynamics. Delphi prioritizes anonymity and iterative refinement, which enhances the reliability of expert consensus but often leads to fatigue in multi-round implementations. By contrast, NGT achieves rapid convergence in a single session and fosters inclusivity, though its lack of anonymity may introduce power asymmetries, particularly in hierarchical project environments. This divergence illustrates a key trade-off between methodological rigor and participatory inclusiveness, both of which must be balanced depending on stakeholder configurations.

Similarly, the two system-oriented techniques—HAZOP and PHA—differ in depth and operational scope. HAZOP is characterized by its high diagnostic capacity, systematically decomposing technical systems to identify compound failures. However, its application requires substantial preparatory documentation and facilitation expertise, limiting its practicality in early design or resource-constrained projects. PHA, in contrast, offers a rapid, checklist-based approach suitable for early-phase hazard scanning; however, its outputs are often generic and require further elaboration before being integrated into detailed planning. Taken together, these findings underscore that while all four methods contribute value, their utility is context-dependent, shaped by trade-offs between depth, efficiency, inclusiveness, and resource intensity. To provide a more intuitive overview of these distinctions,

Table 9. Matrix Summary of Risk Identification Techniques.

Method	Main Strengths	Limitations	Dominant Risk Focus	Overlap with Others
Delphi	Deep expert elicitation; accommodates tacit knowledge; systematic consensus-building	Time-consuming; risk of participant fatigue; requires experienced facilitators	Contextual/PM risks (budget underestimation, regulatory delays, design–construction misalignment); some OHS (hazardous excavation, seismic risks)	Overlaps with PHA (permits, site planning) and NGT (stakeholder conflicts)
NGT	Highly participatory; inclusive; transparent prioritization	Lack of anonymity; potential social desirability bias in hierarchical settings	OHS risks (crane accidents, scaffolding failures, equipment breakdowns, fatigue) and contextual risks (logistics bottlenecks, supplier insolvency)	Overlaps with Delphi (stakeholder conflicts) and HAZOP (dust exposure, utility strikes)
HAZOP	Rigorous, system-based; detailed node analysis; strong for technical hazards	Resource-intensive; requires detailed documentation; not suited for early project phases	OHS risks (incorrect reinforcement, electrical hazards, ventilation issues, equipment malfunctions)	Overlaps with Delphi (utility conflicts, hazardous excavation) and PHA (early excavation collapse, material exposure)
PHA	Lightweight, rapid screening; efficient for early stages	Generalized outputs; lacks detail for task-level planning	Early-stage OHS risks (falls from height, emergency response gaps, material mismanagement) and contextual risks (permits, site planning, environmental impacts)	Overlaps with Delphi (permits, site planning) and HAZOP (excavation collapse, dust/noise exposures)

presents a comparative matrix that summarizes the performance of Delphi, NGT, HAZOP, and PHA across the evaluative criteria (C1–C7). This tabular visualization complements the narrative synthesis by allowing readers to quickly discern methodological differences, overlaps, and trade-offs, thereby enhancing the accessibility and practical utility of the comparative evaluation.

5.4. Cognitive and Behavioral Dimensions of Method Deployment

In addition to structural and technical attributes, the cognitive and behavioral dynamics associated with each technique emerged as influential factors during the applications. These dimensions—often overlooked in formal evaluations—became particularly salient in methods involving real-time group interaction.

For instance, the NGT workshop engaged participants—including site engineers and crew leaders—in identifying site-specific risks. However, the lack of anonymity meant that in the presence of senior personnel, some junior participants self-censored, reflecting known concerns about social desirability bias and hierarchical influence. The literature indicates that NGT, despite its structured format, does not fully mitigate interpersonal and power-related pressures in group settings (e.g., [60]). Moreover, broader research on social desirability bias underscores how social hierarchies can systematically suppress open expression [61]. Thus, while NGT’s stepwise procedure facilitates inclusive participation, organizational culture and interpersonal relationships can still moderate its effectiveness [60,61].

In contrast, the Delphi method, implemented through asynchronous and anonymized expert rounds, reduced interpersonal pressure and encouraged nuanced, experience-based risk articulation. Delphi’s anonymity is well-documented to reduce dominance and conformity effects, promoting more honest and reflective responses [62]. However, its iterative and remote nature introduces limitations: the lack of real-time feedback slows responsiveness, extends timelines, and can cause expert fatigue or attrition [60,63]. Additionally, specific operational nuances—especially those rooted in direct site exposure—were identifiable in the NGT but not captured by Delphi, likely due to its abstraction from the context.

These findings underscore the importance of method selection that considers not only formal structure but also the project’s psychosocial fabric. Real-time collaborative techniques, such as NGT, must be deployed with awareness of hierarchy and communication norms. In contrast, remote or anonymized methods, like Delphi, must address the risks of disengagement or abstraction. Ultimately, no technique is behaviorally neutral; each interacts with the cognitive and relational dynamics of its users, influencing both the risk-identification process and its outcomes.

5.5. Integration with Broader Risk Management Architecture

Another key insight from the case study concerns the integration of identified risks into the broader project risk management workflow. The application of each method revealed critical differences in how their outputs support—or hinder—subsequent phases such as risk analysis, prioritization, mitigation planning, and continuous monitoring.

For example, HAZOP sessions generate structured, scenario-based risk statements tied to specific system functions, which are widely recognized for their compatibility with fault tree models and probabilistic risk assessments [55,64]. This facilitated direct integration into the project’s risk register and supported downstream quantitative analysis. However, the extensive preparation and documentation workload imposed practical constraints, especially in fast-paced project environments with limited staff resources [26].

In contrast, PHA produced general risk categories that proved useful for early-stage screening but lacked the granularity necessary for detailed mitigation planning. Literature on PHA highlights its role as a broad scanning tool, requiring subsequent refinement before it can be translated into actionable safety measures [65,66].

The NGT workshop, while rich in operational detail and inclusive of diverse frontline perspectives, yielded context-specific qualitative insights without a system-level structure. These results were valuable for immediate corrective actions but required reformatting to align with formal reporting or compliance tools [34].

Similarly, Delphi produced a broad set of expert-informed risks, but due to the absence of standardized reporting templates, the outputs varied in format and depth. This is consistent with methodological critiques of the Delphi method, which emphasize variability in outcomes and the additional effort required for synthesis and prioritization [31,60].

These patterns reinforce the consensus in risk management scholarship that risk identification should not be isolated from the broader project risk management architecture. A technique’s operational value lies not only in the risks it reveals but also in the usability, traceability, and interoperability of those insights [21,50,67]. Consequently, the selection of a method must take into account the intended trajectory of risk information throughout the project lifecycle. Methods that provide structured, high-fidelity data may be better suited to projects with robust analytical infrastructures, whereas flexible or participatory methods may require organizational adaptations to ensure continuity across risk management stages.

5.6. Empirical Validation Through Case Study

The contextual decision-support model developed in Section 5.10—Designed to guide method selection based on project attributes such as technical complexity and stakeholder dynamics—was retrospectively validated against the empirical findings from the risk identification exercises conducted within the Water Treatment Plant project. The application of the techniques provided a unique opportunity to assess whether the model’s decision pathways aligned with real-world performance under field conditions.

Specifically, the model recommended the HAZOP method for high-complexity project segments that required detailed technical interrogation and formal system analysis. This recommendation was borne out during implementation: HAZOP proved highly effective in surfacing latent failure scenarios within mechanical and geotechnical subsystems, consistent with its predicted evaluative advantages under criteria C1, C3, and C6. This observation aligns with prior studies that highlight HAZOP’s strength in systematic hazard detection and its suitability for integration with fault tree and probabilistic risk assessments [55,64]. However, its intensive resource requirements also confirmed the model’s caution that HAZOP is only appropriate when documentation is mature and specialist personnel are available [26].

In parallel, the Delphi method was suggested by the model as suitable for scenarios demanding structured expert input without real-time interaction—such as regulatory ambiguity or long-term strategic risk. Its actual deployment corroborated this: Delphi revealed system-level vulnerabilities and institutional coordination gaps, albeit with engagement fatigue and time delays, as anticipated by the model’s logic under C4 and C2. These outcomes are consistent with the methodological literature, which documents both the advantages of anonymity and iteration, as well as the drawbacks of time consumption and participant attrition [31,60].

The model’s routing toward NGT in stakeholder-sensitive, operationally dynamic contexts also aligned with practice. The NGT session captured frontline risks and fostered team ownership, especially among mid-tier site personnel. Although the model flagged potential issues around depth (C3) and anonymity (C2), these were observed in moderated form, reinforcing the model’s diagnostic value. This finding aligns with research indicating that NGT offers inclusive and context-specific operational insights, but is limited by group dynamics and a lack of anonymity [34,57].

Finally, PHA was correctly predicted to be beneficial for early-stage or time-limited conditions where rapid, low-resource risk scanning was needed. Its empirical performance validated the model’s framing of PHA as an agile but shallow tool, suitable primarily for initial scoping rather than detailed planning. This reflects prior work that recognizes PHA’s role in broad hazard categorization, while emphasizing its limited capacity for detailed mitigation planning [65,66].

Taken together, the empirical alignment between the model’s logic and real-world outcomes strengthens its credibility as a practitioner-oriented tool. While the decision tree simplifies a complex process, its structure proved robust in determining the suitability of forecasting methods across varying site conditions. Future applications across diverse project types and organizational settings will be essential to validate further and refine the model, consistent with broader project risk management frameworks that stress iterative validation and contextual adaptation [21,50,67].

5.7. Comparative Reflections Based on Evaluation Criteria

The comparative implementation of the risk identification techniques within the context of the Drinking Water Treatment Plant project has provided rich empirical insights into their practical performance, contextual alignment, and evaluative differentiation across the seven established criteria (C1–C7). These observations, grounded in actual field dynamics and stakeholder interactions, illuminate the nuanced trade-offs and operational realities associated with each method.

C1 Methodological Structure: Both Delphi and HAZOP exhibited pronounced procedural robustness, but their formal requirements—multi-stage iteration for Delphi and guideword-driven logic for HAZOP—were only feasible due to the project’s advanced design maturity and relatively stable planning horizon. Delphi’s iterative rounds enabled the structured convergence of dispersed expert opinions, particularly regarding regulatory uncertainties and supply chain fragilities, consistent with established accounts of its iterative strength [31,60]. HAZOP’s process flow interrogation facilitated the identification of mechanical interdependencies, such as pump and valve integration, echoing prior findings on its diagnostic rigor [55,64]. However, both formats demanded significant facilitation effort, which literature recognizes as a limitation in smaller-scale or earlier-stage projects [26]. In contrast, NGT and PHA offered lighter procedural architectures, enabling rapid deployment within construction coordination meetings, albeit with compromises in depth [65].

C2 Stakeholder Participation: NGT was implemented through a facilitated workshop with the site engineer, safety supervisor, and four crew leaders, allowing for equitable input across occupational tiers. This inclusive participation revealed context-specific hazards, such as vehicular access conflicts and operator fatigue—an outcome consistent with the literature highlighting NGT’s participatory strengths [34,57]. PHA, while more managerially centered, was sufficiently flexible to include cross-functional contributions. Delphi, by contrast, remained expert-centric, as noted in prior critiques [50], and HAZOP required participants with technical literacy, limiting broader engagement [55].

C3 Analytical Depth: The most substantial yield emerged from the HAZOP and Delphi analyses. HAZOP’s ability to detect compound failure modes (e.g., chemical backflow risks due to valve sequencing errors) is well-documented in safety engineering research [66]. Delphi’s open-ended format elicited systemic risks relating to permitting bottlenecks and cross-agency delays, aligning with prior methodological studies [31]. Both required extensive preparatory work, including up-to-date P&IDs for HAZOP and multiple expert rounds for Delphi. In contrast, NGT and PHA identified immediate operational risks, such as fall hazards, but lacked the capacity to detect latent vulnerabilities, as found in previous evaluations of heuristic techniques [65].

C4 Resource Intensity: Delphi required structured communication rounds over multiple weeks, a limitation consistent with its known time and participation burdens [53]. HAZOP, likewise, was document- and expert-intensive, echoing the constraints outlined by Kletz [64]. These methods could only be applied to selected subsystems under tight mobilization schedules. Conversely, NGT and PHA were implemented within routine coordination sessions and required minimal materials, a strength noted in the literature regarding low-resource contexts [34,65].

C5 Flexibility and Adaptability: NGT and PHA demonstrated strong adaptability. NGT was applied across different site teams, producing actionable yet distinct risk maps—a flexibility consistent with its application in dynamic construction settings [28]. PHA checklists were updated midstream to capture environmental compliance risks, reflecting its recognized agility [65]. In contrast, HAZOP’s rigid structure resisted adaptation, and Delphi’s linear, round-based protocol proved inflexible under evolving timelines [60].

C6 Decision Support Value: Delphi and HAZOP provided the highest integration with project risk registers and planning. Delphi prompted a proactive reassessment of procurement timelines, while HAZOP outputs informed commissioning procedures—confirming their high decision-support capacity [55,68]. NGT produces enriched toolbox talks and field safety checklists, echoing earlier findings on the utility of its operational decision-making [34], although its strategic reach remains limited. PHA offered early scoping but little downstream integration [66].

C7 Contextual Fit: In bureaucratically layered environments with limited documentation, NGT and PHA aligned well with prevailing institutional practices, supporting engagement without requiring heavy procedural infrastructure—an advantage emphasized in prior studies [21,50,67]. Delphi, although rigorous, was constrained by the availability of experts and the tolerance for iteration [60], whereas HAZOP required significant adaptation in documentation and facilitation [64].

In summary, the comparative application of these techniques within the Water Treatment Plant project demonstrated that method suitability is highly context-contingent. Structured, resource-intensive techniques like Delphi and HAZOP offer deep insights but require institutional maturity, whereas flexible and participatory approaches, such as NGT and PHA, deliver responsiveness under constrained conditions. These findings underscore the necessity of calibrating method selection based on project phase, organizational readiness, and contextual complexity, consistent with the broader risk management scholarship [21,50,67].

5.8. Contextual Decision Support Model for Risk Identification Method Selection

To address the gap between conceptual evaluation and applied method selection, this study proposes a Contextual Decision Support Model grounded in both analytical criteria and real-world project dynamics. Informed by the structured application of risk identification techniques within a complex infrastructure setting, the model synthesizes theoretical rigor with empirical insights to support informed, context-sensitive selection of appropriate methods across diverse project environments [21,50,67]. Figure 4 visualizes the diagnostic pathway introduced above.

The model is designed around key diagnostic checkpoints that reflect observable project attributes such as technical complexity and the need for stakeholder consensus. These checkpoints are explicitly mapped to relevant evaluation criteria. For instance, projects characterized by high technical complexity (e.g., infrastructure, energy) trigger an assessment of whether detailed technical system analysis is required. If so, HAZOP is recommended due to its formalized, scenario-based structure and high diagnostic depth, which is consistent with its established role in identifying compound system failures [55,64]. If not, the Delphi method becomes more suitable, enabling structured expert elicitation with lower documentation requirements and strong potential for synthesizing expert consensus [31,60].

In contrast, low-to-moderate complexity projects (e.g., residential buildings, roadworks) shift the decision logic toward stakeholder dynamics. Where stakeholder consensus is a priority—such as in participatory design or community-driven development—the model points toward the NGT, widely recognized for its inclusiveness and transparency in group decision-making [28,47]. If consensus is not essential, PHA is recommended as a rapid, resource-efficient screening tool suitable for early-stage hazard scanning [65,66].

Each branch of the decision tree is annotated with the evaluative criteria it reflects, enabling users to trace not only the recommended method but also the rationale behind that recommendation. This approach fosters diagnostic transparency, allowing users to understand which methodological features are prioritized under which conditions, thereby supporting adaptive reasoning rather than rigid prescription.

Importantly, the empirical findings from the case study further validate the contextual logic embedded in the model. For instance, while HAZOP proved highly effective in identifying latent technical risks, its resource intensity and documentation demands posed significant challenges during field implementation, reinforcing its suitability only for high-complexity projects with adequate preparatory resources [26]. Delphi, although methodologically robust, encountered constraints related to expert availability and iteration fatigue, suggesting its application is best suited to contexts where asynchronous expert input is feasible [63]. NGT emerged as a highly accessible technique that promoted worker engagement and ownership, though its outputs remained operational and lacked system-level abstraction, confirming its relevance in low-complexity projects with strong site-level participation [34]. PHA, while efficient and easy to deploy, produced generalized insights that required subsequent translation into actionable items—aligning with prior assessments of its role as an early-stage hazard identification tool rather than a detailed planning method [65,66].

Taken together, these empirical patterns corroborate the model’s bifurcated logic and strengthen its practical credibility. Rather than abstract theorization, the model reflects lived project conditions and method performance in action. While the proposed model offers a valuable aid for strategic method selection, it is not intended as a one-size-fits-all tool. Its utility depends on the accurate assessment of project attributes and practitioner judgment. In ambiguous or hybrid projects, where technical, organizational, and social complexities overlap, professional discretion and iterative deliberation remain essential [67]. Nevertheless, by aligning evaluative reasoning with contextual awareness, the Contextual Decision Tree for Risk Identification Method Selection provides a rigorously grounded and practically practical framework for improving methodological alignment in construction risk management. In this sense, the model not only facilitates informed method selection but also cultivates reflective practice in construction risk governance.

5.8.1. Worked Examples: Application of the Decision Model

To illustrate the practical functioning of the decision tree, two detailed scenarios are presented, each highlighting how project-specific characteristics lead to the final selection of an appropriate risk identification method.

The first scenario involves the construction of a metropolitan subway tunnel beneath a densely populated urban district. The project is part of a strategic transport program designed to alleviate congestion in a rapidly growing city. Early investigations reveal uncertain soil stratigraphy, potential water-bearing layers, and proximity to existing underground utilities. These features indicate high technical complexity and significant safety risks. Applying the decision tree, the evaluator first categorizes the project under “high complexity with strong safety-critical components.” This directs the path toward a system-based analytical technique. After following the decision nodes, the final recommendation is HAZOP, as it systematically decomposes tunneling operations into discrete components (e.g., excavation shafts, lining installation, mechanical ventilation, and conveyor systems). Standard guidewords are applied to each node, producing deviations such as “unexpected groundwater inflow,” “insufficient ventilation in confined areas,” and “formwork instability.” For each deviation, the causes, consequences, and safeguards are documented, resulting in a structured hazard register. Final decision: HAZOP is selected because it best matches the project’s technical depth and safety-critical environment.

The second scenario concerns a large-scale social housing development commissioned by a municipal government. Unlike the tunneling project, the housing program involves the repetitive construction of multi-story residential blocks over two years. While the technical elements are relatively standardized, the project faces heightened social sensitivity. Residents express concerns about noise, dust, and traffic disruption; contractors are under pressure to meet tight delivery deadlines; and municipal agencies require stakeholder-inclusive processes. In this context, the decision tree highlights “stakeholder complexity and social sensitivity” as the dominant determinants. This path leads to the branch for group-based consensus. Following the decision nodes, the final recommendation is NGT, since it facilitates structured and inclusive workshops. In practice, six participants (one site engineer, one safety supervisor, and four crew leaders) identified and prioritized 25 distinct risks, including “dust exposure affecting nearby schools” and “traffic congestion from construction logistics.” Through voting and ranking, dust exposure and traffic congestion emerged as top priorities. Final decision: NGT is selected because it provides a participatory, consensus-driven structure that directly addresses stakeholder-related concerns.

Together, these scenarios confirm that the decision tree moves beyond abstract evaluative criteria by producing clear, context-specific outcomes. The subway tunnel case demonstrates that HAZOP is the optimal choice for technically complex and safety-critical settings, while the housing project illustrates that NGT is most effective in socially sensitive and stakeholder-intensive environments. The explicit inclusion of these final decisions demonstrates how the decision tree not only guides the reasoning process but also delivers actionable outputs for practitioners.

5.8.2. Application Pathways: Owners, Contractors and Insurance Companies

While the Contextual Decision Support Model provides a transparent framework for selecting appropriate risk identification techniques, its value ultimately depends on how it can be operationalized by different stakeholders in construction projects. Owners, contractors, and insurance companies approach risk from distinct vantage points: owners emphasize strategic governance and accountability, contractors focus on execution and operational safety, and insurers evaluate the credibility of risk practices to inform financial coverage decisions. By tailoring the decision tree’s diagnostic logic to these different needs, the model offers concrete implementation pathways that bridge methodological guidance with real-world practice.

For project owners, the Contextual Decision Support Model provides a structured tool for aligning risk identification practices with overarching strategic goals. Owners typically prioritize financial predictability, timely delivery, regulatory compliance, and stakeholder legitimacy [69]. The decision tree allows owners to select techniques that best address these concerns at the earliest stages of procurement and planning. For instance, in large-scale, technically complex infrastructure projects, the model directs owners toward HAZOP or Delphi, ensuring that systemic and latent risks (e.g., geotechnical variability, utility conflicts) are systematically assessed before investment decisions are finalized. In contrast, in socially sensitive or community-driven developments, the model emphasizes the suitability of NGT, allowing owners to incorporate local concerns, such as noise, dust, and disruption, into their risk registers. By embedding these method-specific requirements into tender specifications or contract clauses, owners can ensure that risk identification is not left to contractor discretion but becomes a contractual deliverable. This approach aligns with emerging trends in risk-based contracting [70], where method selection is increasingly tied to project governance and accountability.

Contractors must balance technical execution with cost, time, and safety constraints, making efficient and adaptive risk management essential. The decision tree can guide contractors in sequencing risk identification methods in alignment with project phases and available resources. In early stages, contractors may employ PHA for rapid hazard screening (e.g., site access constraints, preliminary excavation risks), providing a foundation for developing early mitigation measures. As detailed designs progress, Delphi can be used to consolidate expert judgments across engineering and subcontractor teams, supporting the refinement of contingency plans. During execution, NGT workshops with frontline supervisors and workers enable the participatory identification of operational risks, such as crane operation hazards, scaffolding instability, or equipment breakdowns. This phased approach helps contractors allocate resources efficiently, reserving high-intensity methods (e.g., HAZOP) for the most critical technical systems, while relying on lighter tools for operational coverage. Evidence from the construction safety literature suggests that participatory techniques, such as NGT, can improve hazard recognition, worker buy-in, and compliance [34,71]. Thus, the model not only enhances methodological rigor but also fosters a safety culture by integrating worker perspectives directly into structured processes.

Insurance providers assess projects not only in terms of financial exposure but also by evaluating the credibility of risk management practices employed by project stakeholders. The decision tree provides insurers with a transparent benchmark for determining whether projects are applying methods proportionate to their technical and contextual complexity. For high-complexity, safety-critical projects, insurers may require evidence that structured analytical techniques such as HAZOP or Delphi have been deployed, thereby reducing the likelihood of latent systemic failures. Conversely, for projects characterized by stakeholder intensity, adoption of NGT is a sign of proactive governance and dispute minimization. This alignment enables insurers to refine their risk pricing models, differentiating premiums based on the demonstrable rigor of their risk identification practices. Similar approaches have been observed in risk-based insurance frameworks where proactive safety and risk management practices result in more favorable premium adjustments [72]. The model thus creates a bridge between technical method selection and financial risk governance, providing insurers with both transparency and assurance in their underwriting decisions.

5.8.3. Applicability in Developed and Global Construction Contexts

While this study is rooted in the context of a developing country infrastructure project, the applicability of the findings can extend beyond such environments when carefully contextualized. Prior research has shown that project governance structures and risk priorities often differ between developing and developed country settings [69]. For instance, while financial predictability and regulatory capacity may dominate in developed economies, institutional fragmentation and resource constraints often shape risk management in developing contexts. Similarly, Zou and Couani [72] emphasize that supply chain vulnerabilities manifest differently depending on local regulatory maturity and industrial capability, highlighting the importance of tailoring risk identification methods to the institutional environment. In this sense, the Contextual Decision Support Model is not intended as a geographically bounded tool, but as a flexible framework that can be adapted to both mature and emerging construction markets.

At the same time, advances in developed country practice provide complementary insights for interpreting and extending the model. Khodabakhshian et al. [73] highlight the growing importance of deterministic–probabilistic integration in risk management for complex projects, suggesting that hybrid frameworks are increasingly valued in high-capacity environments. Taroun [74] similarly points to the limitations of traditional probabilistic models, calling for more context-sensitive approaches that acknowledge multidimensional project dynamics, a principle that underpins the present model as well. More recently, Yazdi et al. [75] have demonstrated the growing role of AI-driven and machine learning–based systems in risk management, which can augment expert-based approaches by providing predictive insights under uncertainty. These contributions suggest that while structured expert techniques remain foundational, their integration with simulation, BIM-enabled risk analytics, and AI-based platforms represent a promising frontier for global practice.

In multinational and global construction companies, the utility of a decision-support framework lies in its ability to harmonize risk identification across projects with widely varying contexts. Zhang et al. [76] argue that knowledge management systems can enhance performance by standardizing risk-related practices across organizational units, a goal closely aligned with embedding diagnostic criteria such as those in the decision tree. Moreover, Lingard et al. [71] emphasize that occupational health and safety risks demand a systemic perspective even in advanced economies, underscoring the continued relevance of structured participatory tools such as NGT. In this light, the proposed model provides a scaffold for global firms to calibrate their method selection according to project complexity, stakeholder dynamics, and cultural environment, while remaining compatible with emerging digital and AI-based systems.

Taken together, these insights reinforce that the decision-support framework developed in this study is not confined to the challenges of developing country settings. Instead, it contributes to the broader agenda of improving risk governance in global construction, offering a flexible foundation that can be tailored to diverse project environments and integrated with cutting-edge digital technologies.

5.8.4. Positioning the Study Within Decision-Support Systems for Construction Risk Management

The decision-support tree developed in this study can also be situated within the broader stream of research on decision-support systems for construction risk management. For example, Dikmen et al. [77] introduced a decision-support tool that integrates risk and complexity assessment through visualization, enabling practitioners to understand project dynamics better. Similarly, Taroun [74] highlighted the limitations of traditional probabilistic models and called for more context-sensitive approaches that capture the multidimensional nature of construction risk. More recently, Khodabakhshian et al. [73] compared deterministic and probabilistic strategies, highlighting the need for hybrid frameworks that strike a balance between methodological rigor and practical applicability. While these contributions provide valuable insights into risk assessment and visualization, the present study emphasizes a different but complementary dimension: guiding the selection of risk identification techniques according to project context and stakeholder needs. In addition to these traditional and hybrid decision-support approaches, Yazdi et al. [75] demonstrated how artificial intelligence can be leveraged in risk management through comparative analysis of AI-driven tools. Their work highlights the growing role of machine learning and intelligent systems in supporting decision-making under uncertainty. While such AI-based frameworks aim to enhance predictive capacity, the present study complements this stream by focusing on the operational selection of risk identification techniques, offering a transparent and context-sensitive decision tree that can be integrated with future AI-driven platforms. Taken together, these approaches illustrate the evolving landscape of risk-informed decision-support tools in construction, where both quantitative modeling and qualitative method-selection frameworks play critical and mutually reinforcing roles.

5.9. Comparative Contextual Evaluation of Risk Identification Techniques

Table 10. Comparative Evaluation of Risk Identification Techniques Across Multi-Criteria Dimensions.

Criterion	Delphi Method	NGT	HAZOP	PHA
C1. Methodological Structure	(+) Highly structured, systematic, and iterative process (−) Long implementation time	(+) Clear procedural steps, easy to implement (−) Lack of anonymity may create conformity pressure	(+) Highly detailed and technically robust procedure (−) Limited applicability at early project stages	(+) Simple and quick structure (−) Does not provide in-depth analysis
C2. Stakeholder Participation	(+) Deepens expert insights through multiple rounds (−) Limited participant interaction (no face-to-face)	(+) Encourages active and equal participation (−) Diversity of views may decline in high power-distance contexts	(+) Intensive interaction among multidisciplinary teams (−) Limited contribution from non-experts	(+) Can quickly gather input from different stakeholder groups (−) Engagement may remain superficial
C3. Analytical Depth	(+) Enables deep analysis and synthesis of ideas (−) Requires substantial time and data	(+) Provides moderate analytical depth in practice (−) Limited systematic technical analysis	(+) Highest level of technical analytical depth (−) Cannot be applied without sufficient documentation and data	(+) Conducts basic risk screening (−) Does not offer detailed analysis
C4. Resource Intensity	(−) High demand for time, coordination, and human resources	(+) Low cost and time requirements	(−) High technical expertise and preparation needed	(+) Lowest resource requirements
C5. Flexibility & Adaptability	(−) Rigid process, difficult to change (+) Adjustable by varying the number of rounds	(+) Easily adaptable to different project types	(−) High dependence on documentation	(+) Quickly integrates into early-stage projects
C6. Decision Support Value	(+) Strong, data-driven insights for decision-making	(±) Strong for short-term decisions, weaker for long-term	(+) Directly linkable to risk mitigation planning	(−) Weak decision support, requires additional analysis
C7. Contextual Fit	(±) Strong where expert access is easy and there is no time pressure (−) Challenging under time constraints or with a limited expert pool	(+) Ideal for environments with low documentation and rapid decision needs	(±) Effective in projects with high documentation and technical maturity (−) Difficult to apply in developing sectors	(+) Suitable for contexts with limited resources and quick decision-making needs

(+) denotes a strength or advantage of the method in relation to the given criterion; (−) denotes a weakness or limitation; (±) indicates a context-dependent or neutral effect, where the outcome varies according to project-specific conditions.

provides a comprehensive comparative evaluation of the four risk identification techniques—Delphi, NGT, HAZOP, and PHA—across the seven predefined criteria (C1–C7) established in this study. The matrix is not intended merely as a static summary of strengths (“+”), weaknesses (“−”), and context-dependent outcomes (“±”), but as an interpretive framework that underscores the situational fit of each method. By systematically aligning methodological attributes with project-specific demands, the table functions as a structured decision-support tool that enables practitioners to identify the most contextually appropriate technique rather than assuming a universally superior option [21,37,67].

The results reveal distinct methodological profiles. Delphi stands out for its strong methodological structure (C1) and analytical depth (C3), producing high-quality insights through iterative expert consultation [25,50]. It also demonstrates significant decision-support capability (C6), particularly in complex or strategic settings [68]. However, these benefits are offset by high resource intensity (C4), as Delphi requires considerable time, coordination, and expert availability, and its contextual fit (C7) diminishes in projects that demand rapid outputs [63].

HAZOP similarly excels in its systematic structure (C1) and delivers the highest levels of technical analytical depth (C3) and decision-support relevance (C6), making it highly reliable in safety-critical environments [45,54]. Yet, its dependence on detailed documentation, expert facilitation, and preparatory resources makes it less adaptable in early-stage or resource-constrained projects [26].

In contrast, NGT demonstrates its most significant strengths in stakeholder engagement (C2) and flexibility (C5), encouraging inclusive participation and adaptability across diverse project types. This makes it particularly well-suited for contexts requiring legitimacy and collaborative input [34,57]. However, NGT offers only moderate analytical depth (C3), and the lack of anonymity can expose the process to conformity pressures, particularly in hierarchical organizational cultures [61].

PHA, on the other hand, is characterized by its simplicity and low resource requirements (C4), making it a practical choice for preliminary screening or projects with tight time and budget constraints [65,66]. Its outputs are rapid and adaptable (C5, C7), but its analytical rigor (C3) and decision-support capacity (C6) remain limited, restricting its use in complex, high-stakes environments.

A cross-analysis of the seven criteria reveals essential trade-offs. The balance between analytical depth (C3) and resource demands (C4) is most visible in Delphi and HAZOP, whereas the tension between stakeholder inclusivity (C2) and decision-support robustness (C6) distinguishes NGT and PHA. These interdependencies highlight that evaluation criteria cannot be considered in isolation; instead, they must be understood as part of a broader methodological ecosystem [21,37,67].

In sum, Table 4 transcends mere tabular comparison by operationalizing contextual alignment. Each technique offers a unique balance of strengths and limitations, reinforcing that method selection should be criteria-driven and project-specific. By capturing the interplay between analytical depth, stakeholder dynamics, resource availability, and contextual maturity, the matrix equips practitioners with a transparent, evidence-based framework to enhance the rigor, adaptability, and robustness of risk management strategies in construction projects.

5.10. Radar Chart Visualization of Comparative Performance Across Risk Identification Techniques

To complement the tabular comparison, a radar chart was developed to visualize the relative performance of the four techniques across the seven evaluation criteria (C1–C7). As shown in Figure 5, this visualization highlights not only the numerical trends but also provides an integrated textual interpretation. For instance, Delphi demonstrates strength in methodological structure and analytical depth, consistent with its iterative and systematic expert-driven design [31,60]. However, it is resource-intensive and rigid in practice, as noted in prior critiques of time and participation burdens [63]. NGT appears as a more balanced approach, characterized by high stakeholder engagement and flexibility, which echoes its recognized ability to facilitate inclusive participation across diverse groups [34,57]. However, it lacks systematic analytical rigor due to its limited depth and the absence of anonymity. HAZOP stands out for its highest technical robustness and decision-support capability, aligning with its well-documented capacity for uncovering complex system failures in process industries [55,64]. However, it is heavily dependent on documentation and expert availability, which constrains its application in early-stage or resource-limited contexts. In contrast, PHA offers quick implementation and minimal resource requirements, consistent with its role as a preliminary hazard screening method [65,66]; yet, its analytical depth and decision-support potential remain limited. This dual-layer presentation (visual and textual) enables a holistic understanding of the strengths and weaknesses of each technique, facilitating more explicit comparisons and context-sensitive decision-making.

5.11. Heatmap Visualization of Technique Performance by Evaluation Criteria

To provide an intuitive and comparative overview of how each risk identification technique performs across multiple evaluation criteria, a heatmap visualization was constructed (Figure 6). This tool translates the weighted scores of Delphi, NGT, HAZOP, and PHA into a color-coded intensity map, where darker shades represent higher suitability and lighter shades indicate weaker performance.

The heatmap enables quick pattern recognition, allowing for the identification of techniques that consistently perform well across diverse dimensions and where trade-offs occur. For instance, Delphi demonstrates vigorous intensity in areas of stakeholder engagement and decision support, consistent with its recognized capacity for synthesizing expert judgment through structured rounds [31,60]. However, it exhibits lighter performance in resource efficiency, reflecting the known participation and time demands [63]. NGT presents balanced coloration, with moderate to vigorous intensity in efficiency and adaptability, echoing its effectiveness in inclusive, practice-based engagement [34,57], although with less emphasis on analytical depth due to its qualitative focus. HAZOP, in contrast, exhibits a concentrated intensity in technical rigor and safety-critical contexts, aligning with its established role in uncovering latent system hazards [55,64], while appearing to have lighter resource demands and a higher inclusivity of stakeholders [26]. Finally, PHA demonstrates intense coloration in early-stage applicability and flexibility, consistent with its use as a rapid screening tool [65,66], but shows weaker tones in structured depth and long-term decision support.

By visualizing the comparative scores in a heatmap, the analysis goes beyond numerical averages and highlights clusters of strengths and weaknesses across techniques. This approach provides a holistic perspective for practitioners, enabling them to quickly assess which technique best aligns with the dominant priorities of their project environment.

5.12. Summary and Implications

In summary, the comparative assessment of the four structured risk identification techniques—Delphi, NGT, HAZOP, and PHA—demonstrates that their practical effectiveness in construction projects is primarily shaped by contextual determinants rather than methodological hierarchy. While HAZOP exhibited the most substantial analytical depth and decision-support capability, NGT proved most effective in fostering participatory stakeholder engagement under time- and resource-constrained conditions. Delphi provided robust systemic insights, yet required significant expert availability, whereas PHA enabled rapid screening but offered limited analytical depth. These findings reinforce prior studies (e.g., [26,34,72]) that emphasize the value of structured and participatory approaches, yet they extend the literature by empirically illustrating how contextual alignment—project complexity, stakeholder diversity, and institutional maturity—ultimately governs the applicability.

Methodologically, the study advances a transparent multi-criteria evaluation framework (C1–C7) that integrates empirical evidence with international standards (ISO 31000 [50] and PMBOK Guide 7th Ed. [21]), thereby bridging conceptual models and field implementation. Practically, it delivers a contextual decision-support model, complemented by radar and heat-map visualizations, that helps practitioners balance methodological rigor with feasibility under real-world constraints. Collectively, these contributions provide a holistic and actionable understanding of how structured risk identification can be systematically integrated into construction project management, thereby enhancing both predictive insight and organizational learning.

6. Research Limitations and Recommendations for Further Research

6.1. Limitations of Study

Despite the methodological contributions of this study, several limitations should be acknowledged. First, the expert panels employed in the Delphi (n = 5), NGT (n = 6), HAZOP (n = 4), and PHA (n = 2) applications were relatively modest in size. While such sample sizes are consistent with methodological guidance for structured expert techniques—where depth of expertise is prioritized over breadth [31,55,78,79,80]—they inevitably limit the generalizability of the findings. The comparative insights derived from the study should therefore be interpreted as illustrative rather than statistically representative. Nonetheless, the use of strict inclusion criteria (a minimum of ten years of professional experience, prior involvement in formal risk management strategies, and role-specific expertise) helped enhance the credibility and stability of the consensus within the panels.

Second, the participant pool was intentionally restricted to technical personnel, including project managers, safety specialists, design engineers, and site supervisors. Non-technical stakeholders, such as clients, policymakers, residents, and operators, were not included in the structured exercises. This choice was guided by the study’s primary objective, which was to evaluate the operational strengths, challenges, and trade-offs of the selected risk identification techniques, rather than to produce a fully comprehensive inventory of project risks. As a result, potential biases linked to the exclusion of non-technical perspectives must be acknowledged. Future research could address this limitation by expanding participant diversity to include a broader set of stakeholders, thereby capturing the social and institutional dimensions of risk perception more comprehensively.

Third, this study deliberately adopted an interpretivist stance and therefore did not assign numerical weights to the evaluative criteria (C1–C7). While this qualitative orientation was selected to preserve methodological nuance and contextual richness, the criteria set was intentionally designed to be compatible with quantitative extensions. In future research, these dimensions could be operationalized using multi-criteria decision-making (MCDM) techniques, such as the Analytical Hierarchy Process (AHP) or Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS), which enable weighted comparisons and numerical prioritization of methods. Such extensions were beyond the scope of the present study but represent a promising avenue for integrating interpretivist insights with quantitative decision-support. While no numerical scoring system was applied, the safeguards employed (double-coding, peer review, triangulation) provide methodological robustness and replicability. Future research may extend this framework by utilizing multi-criteria decision-making tools, such as AHP or TOPSIS, to the same criteria set.

Fourth, while the present study captures a static application of four techniques within a single project phase (detailed planning), it is essential to recognize that risk identification is inherently dynamic and iterative [50]. As projects evolve from conceptual design to execution and operation, both the availability of information and the nature of risks change [74]. In early stages, methods such as PHA are particularly suitable for broad hazard scanning under conditions of uncertainty, whereas Delphi may later support structured expert consensus once more technical and organizational data become available. During execution, participatory techniques such as NGT can be valuable for capturing site-based and operational risks [34], while HAZOP is best reserved for phases with mature design documentation and system-level detail. This adaptive sequencing highlights that the choice of technique should not be static, but rather evolve in response to project complexity, stakeholder dynamics, and data maturity. Future research could explicitly investigate phase-based application of risk identification methods, exploring how hybrid or sequential deployment across the project lifecycle can enhance both efficiency and comprehensiveness of risk management.

Finally, the study focused exclusively on four structured methods (Delphi, NGT, HAZOP, and PHA). While these represent established and widely documented techniques, other approaches, including fault tree analysis, bow-tie analysis, fuzzy logic methods, and AI-based risk assessment tools, could also be considered in future work. The exclusion of these alternatives reflects a deliberate scope decision to enable in-depth comparison of the selected techniques, but it also suggests that the findings should be interpreted within this defined methodological boundary.

Taken together, these limitations should not be viewed as diminishing the study’s central contribution. Instead, they provide essential context for interpreting the findings, which remain valuable in advancing comparative, context-sensitive understandings of risk identification methods. At the same time, they highlight fruitful avenues for future research, particularly in terms of incorporating broader stakeholder perspectives and exploring additional methodological approaches.

6.2. Directions for Future Research

6.2.1. Hybrid and Stepwise Applications of Risk Identification Techniques

An additional dimension worth considering is the potential for hybrid or stepwise applications of risk identification techniques. Rather than treating Delphi, NGT, HAZOP, and PHA as entirely separate approaches, future research could explore how they may be combined sequentially or complementarily to address different stages of a project. For instance, PHA can serve as an early-stage screening tool to capture broad categories of hazards, with subsequent HAZOP sessions providing rigorous system-level analysis during the design stage. Similarly, Delphi may be employed to elicit and refine expert insights in contexts characterized by uncertainty, while an NGT workshop could later validate and prioritize risks in a participatory, site-based setting.

The complementarity of these methods is one of their key strengths. HAZOP’s analytical depth can balance PHA’s rapid coverage, while NGT’s inclusiveness and transparency can enrich Delphi’s structured expert consensus. Hybrid pathways, therefore, offer a more holistic and context-sensitive approach, particularly valuable in complex or resource-constrained construction environments. They also create opportunities for efficiency by ensuring that resource-intensive techniques, such as HAZOP, are reserved for areas where their added value is most excellent, while lighter methods, such as PHA or NGT, provide rapid but reliable coverage elsewhere.

Stepwise integration enhances adaptability across project phases. Risks identified through the PHA in the conceptual stage can inform Delphi consultations during detailed design, with the results later validated through NGT in stakeholder coordination meetings. This would transform risk identification from a one-off task into a dynamic, iterative process aligned with evolving project needs. Such an approach resonates with contemporary calls for flexible, context-driven risk governance that moves beyond static or single-method applications.

Evidence from related fields reinforces the value of such integration. In process safety research, PHAs have been combined with HAZOP to focus resources on the most critical system nodes [81,82]. In organizational domains, sequencing Delphi and NGT has been shown to enhance both methodological rigor and inclusiveness by balancing expert-driven refinement with participatory consensus [33]. Building on these precedents, construction risk management could benefit from more systematic exploration of hybrid pipelines—for example, beginning with PHA for broad hazard scanning, followed by Delphi for expert consolidation, NGT for participatory prioritization, and HAZOP for detailed technical analysis.

Although such integrated deployment was beyond the scope of the present study, acknowledging these possibilities highlights a promising avenue for future research and practice. Exploring hybrid and stepwise pathways may not only strengthen methodological robustness but also increase the practical relevance and adaptability of risk identification in diverse construction project contexts.

6.2.2. Emerging Digital Approaches to Risk Identification

While the present study has deliberately focused on structured expert-based techniques (Delphi, NGT, HAZOP, and PHA), it is essential to acknowledge that recent scholarship has increasingly highlighted the role of digital and data-driven approaches in construction risk management. Tools such as Building Information Modeling (BIM)-based risk analysis, artificial intelligence (AI) and machine learning prediction models, and simulation-based techniques are rapidly advancing and are being integrated into industry practice. BIM-based approaches enable the visualization and coordination of risks embedded within design and construction processes, offering dynamic clash detection and scenario planning [83]. AI and machine learning methods, by contrast, provide predictive capacity by processing large-scale project and safety datasets to forecast risk likelihood and severity with increasing accuracy [75]. Simulation tools, including Monte Carlo and system dynamics, further enhance the ability to test probabilistic scenarios and resource allocations under uncertainty [73,74].

These emerging techniques should not be seen as replacements for structured expert-based approaches, but rather as complementary tools. Whereas Delphi, NGT, HAZOP, and PHA provide systematic frameworks for eliciting tacit expert knowledge and fostering participatory consensus, digital methods excel in data processing, visualization, and predictive analytics. A promising direction for future research, therefore, lies in exploring hybrid and integrated models—where expert-driven methods are combined with BIM, AI, and simulation frameworks to capture both qualitative insights and quantitative predictions. Such integration could enhance the adaptability and robustness of risk identification practices across diverse project contexts, ensuring that methodological rigor is matched with technological innovation.

7. Conclusions

This study offered a systematic and empirically grounded comparison of widely used risk identification techniques—Delphi, NGT, HAZOP, and PHA—within the context of construction project management. By employing a multi-criteria evaluation framework and applying each method to a real-world infrastructure project, the research demonstrated that technique selection should not be viewed as a one-size-fits-all decision, but rather as a context-sensitive design choice shaped by project complexity, stakeholder dynamics, and institutional capacity.

The proposed decision-support model provides both theoretical clarity and practical guidance, enabling project stakeholders to align risk identification practices with broader organizational and environmental constraints. Notably, the study highlights that methodological suitability extends beyond analytical precision to include factors such as cognitive bias, participatory equity, and downstream integration potential.

Ultimately, the findings advocate for a more strategic and adaptive approach to risk identification in construction, one that moves beyond static taxonomies toward dynamic, criteria-based configurations capable of enhancing both procedural legitimacy and project resilience.

7.1. Theoretical Contributions

The contribution of this study extends beyond comparative observations by introducing a contextual decision-support framework that systematically integrates evaluative criteria. This multi-criteria framework advances prior models that tend to emphasize a single dimension, such as analytical depth or stakeholder participation, often neglecting the interplay among these attributes in real-world project settings.

The decision tree model developed in this study (see Figure 2) addresses this gap by aligning method selection not only with project technicalities but also with context-specific constraints such as organizational maturity, budget availability, and stakeholder distribution. The implementation was empirically informed, guiding the structure of each technique within the same infrastructure project. For instance, the decision point favoring HAZOP in high-complexity environments is substantiated by its superior performance in uncovering system-level technical hazards during field trials, albeit at the cost of high resource demand and documentation overhead. Conversely, NGT’s suitability for participatory projects was reinforced by its facilitation of inclusive dialogue among operational staff during the on-site session, even if tempered by hierarchical dynamics.

The comparative evaluation framework, which systematically contrasts Delphi, NGT, HAZOP, and PHA across seven interrelated criteria, offers a significant theoretical advancement in the discourse on construction risk identification. It challenges the long-held assumption of methodological neutrality, arguing instead that risk identification techniques are not universally applicable but are highly contingent upon technical, organizational, and institutional conditions.

This insight reframes risk identification from a mechanistic task to a strategic design problem. Rather than seeking an inherently superior method, the study positions each technique as conditionally optimal, depending on its alignment with project-specific demands. Techniques with high analytical depth and structured outputs (e.g., HAZOP, Delphi) are better suited to large-scale or complex projects, while those emphasizing stakeholder inclusivity and procedural simplicity (e.g., NGT, PHA) may prove more effective in resource-constrained or socially sensitive contexts.

A key theoretical contribution is the formalization of contextual fit (C7) as a meta-criterion—extending beyond technical performance to encompass cultural compatibility, organizational behavior, and decision-making norms. This dimension highlights that the epistemological legitimacy of a method is inseparable from the environment in which it is deployed.

By situating method selection within a broader systems-thinking paradigm, the study further conceptualizes risk identification as an upstream determinant of risk management success. Misalignment at this initial stage can cascade into downstream inefficiencies, suboptimal mitigation strategies, or even strategic failure. As such, the choice of technique becomes a foundational element in shaping project resilience.

Furthermore, the model offers diagnostic transparency: each decision branch is annotated with the specific evaluative criteria being activated, helping users not only identify a method but also understand why it is recommended under given conditions. This positions the framework as both a prescriptive and educational tool—bridging the gap between academic typologies and field-level applicability.

Compared to existing taxonomies of risk identification tools, which offer broad categorizations but limited guidance on method selection, the proposed model delivers actionable pathways grounded in empirical performance. Observed trade-offs shaped its development: for example, the infeasibility of Delphi in fast-track project settings due to expert attrition and scheduling constraints, or the limited downstream integration of PHA, which, although time-efficient, required additional steps to convert its high-level findings into structured mitigation plans.

In essence, the decision model reframes method selection as a strategic design problem, where technique suitability emerges not from abstract classification, but from an alignment between method capabilities and project realities. This approach is particularly valuable in project environments—such as construction—where uncertainty, complexity, and resource limitations co-exist and require adaptive, context-aware decision-making.

7.2. Practical Contributions

From a practitioner’s perspective, this study addresses a critical gap in current project management practice: the frequent reliance on intuition, precedent, or institutional habit in selecting risk identification techniques. Instead, it advocates for a structured, evidence-based approach grounded in seven evaluative criteria (C1–C7), enabling practitioners to align method selection with project characteristics and operational constraints.

The framework empowers decision-makers to avoid common mismatches—such as deploying resource-intensive tools in simple projects or using superficial screening methods in high-risk environments. For example, in environments where financial, technical, or human capital is limited, resource efficiency (C4) becomes a decisive factor. Delphi’s iterative format, while analytically robust, may not be viable under accelerated timelines or limited personnel availability. In contrast, NGT offers a lower operational threshold, making it more adaptable to dynamic stakeholder configurations or on-site decision contexts.

Beyond individual project performance, strategic method selection also supports institutional capacity-building. Iterative use of techniques like Delphi fosters internal expertise, while routinizing participatory methods like NGT can cultivate a culture of shared accountability and risk awareness—particularly relevant in fragmented or multi-tiered project teams.

Moreover, the criteria-based model enhances procedural transparency and traceability. As regulatory authorities, funding agencies, and insurers increasingly demand formalized risk management documentation, adopting a structured rationale for technique selection strengthens compliance and facilitates future standardization.

Ultimately, the findings reframe method selection not as a peripheral task but as a strategic intervention with cascading effects on risk identification, prioritization, and mitigation processes across the project lifecycle.

7.3. Methodological Contributions

Methodologically, this study advances the field in two fundamental ways. First, it departs from traditional single-method investigations by offering a comparative, criteria-based assessment of distinct risk identification techniques. This holistic lens allows practitioners and researchers to weigh trade-offs across a multidimensional space, rather than defaulting to familiar or entrenched practices.

Second, the study moves beyond descriptive typologies by introducing a prescriptive decision-support model grounded in empirical case data and validated through contextual application, unlike prior frameworks that emphasize either analytical depth (e.g., HAZOP in process industries) or participatory inclusion in isolation, the model here integrates multiple performance dimensions—creating a practical tool for navigating complex project environments.

This bridging of theory and practice addresses the long-standing methodological divide in the field, highlighting the persistent disconnect between academic sophistication and practical applicability. By offering a replicable and scalable decision model, this study enhances methodological usability, particularly in settings where formal risk governance is still in its early stages and context-sensitive adaptation is crucial.

In sum, the study not only expands methodological rigor but also democratizes its application—positioning structured risk identification as both a scientific discipline and a practical imperative in contemporary project management.