Occupational Safety and Health Risk Analysis Using an Integrated FMEA-AHP and FTA Approach: A Case Study of the Pension Fund Office Building Project of PT Bank East Nusa Tenggara

The construction industry remains one of the most hazardous sectors worldwide, despite sustained advances in occupational safety regulations, technology, and risk management practices. In 2022, the United States Bureau of Labor Statistics recorded 1,069 fatalities among construction workers, with a fatality rate of 9.6 deaths per 100,000 full-time workers [1]. At the global level, the International Labour Organization estimated that occupational accidents and work-related diseases caused approximately 2.93 million deaths in 2019, an increase of more than 5% from 2015. The Asia-Pacific region accounted for nearly 63% of these deaths [2]. These figures show that occupational safety and health (OSH) failures are not isolated events. They reflect persistent weaknesses in hazard recognition, risk prioritization, supervision, and preventive control, particularly in industries characterized by dynamic work environments and multiple interacting hazards.

Indonesia faces a similar and increasingly serious pattern. BPJS Ketenagakerjaan recorded 221,740 occupational accident cases in 2020 and 234,370 cases in 2021. The number rose sharply to 370,747 cases in 2023 [3, 4]. Construction contributes substantially to this burden because project activities involve work at height, heavy equipment, temporary structures, electrical installations, material movement, and continuously changing site conditions. The study [5] reported that the Indonesian construction sector accounts for more than 30% of national occupational injuries. Yet empirical research on construction safety in Indonesia remains insufficient relative to the scale and complexity of the problem. Existing studies also reveal recurring weaknesses, including an underdeveloped safety culture, inconsistent use of personal protective equipment, inadequate supervision, and limited implementation of formal OSH management systems [5, 6].

These problems require closer examination in regional construction settings. Safety risks do not arise solely from technical hazards. They are also shaped by organizational capacity, workforce competence, regulatory enforcement, project scale, contractor characteristics, and local operating conditions. Construction projects in Kupang City, East Nusa Tenggara, operate within a developing regional context that differs from major Indonesian construction centres in terms of infrastructure, access to skilled labour, institutional capacity, safety resources, and project management practices. These contextual differences may influence both the probability of an accident and the effectiveness of existing control measures. However, empirical evidence on OSH risks in building construction projects in Kupang City remains limited [5]. This lack of location-specific evidence restricts the ability of contractors and project managers to identify priority hazards and design controls that respond to the actual conditions of regional construction projects.

A rigorous assessment framework is therefore needed to distinguish critical risks from routine hazards and to explain why priority accidents occur. Failure Mode and Effects Analysis (FMEA) offers a structured approach for identifying potential failure modes and ranking them through the Risk Priority Number (RPN), which combines severity, occurrence, and detection scores [7, 8]. Despite its practical value, conventional FMEA contains important methodological weaknesses. It assigns equal importance to severity, occurrence, and detection, although their relative significance may differ across projects and accident types. It also allows different combinations of severity, occurrence, and detection to produce identical RPN values. Consequently, risks with different implications may receive the same ranking, reducing the discriminatory power of the assessment [9].

The Analytic Hierarchy Process (AHP) can address part of this limitation by assigning relative weights to risk criteria through systematic pairwise comparisons and expert judgment [10, 11]. Integrating AHP with FMEA produces a Weighted Risk Priority Number (WRPN), which reflects the contextual importance of severity, occurrence, and detection rather than treating them as equally influential [12]. This integration strengthens risk prioritization, particularly when project experts possess relevant contextual knowledge that cannot be captured through unweighted numerical scoring. However, FMEA-AHP remains primarily a prioritization instrument. It identifies which failure modes require attention but does not adequately explain the combination of technical, human, and organizational events that produce those failures.

Fault Tree Analysis (FTA) complements FMEA-AHP by tracing a priority accident from the top event to its underlying basic events through logical AND and OR relationships [13]. FTA therefore shifts the analysis from ranking symptoms to examining causal mechanisms. This distinction is essential because effective risk control depends not only on identifying a high-risk event but also on understanding the conditions that allow it to occur. A risk assessment that stops at ranking may lead to generic controls, whereas root cause analysis can support more targeted interventions related to equipment, procedures, supervision, worker behaviour, and organizational systems.

Previous research supports the value of integrating complementary risk assessment methods. An integrated fuzzy FMEA-FTA-AHP model developed for the steel industry demonstrated the analytical value of combining risk weighting with causal analysis [14]. Similarly, integrating FMEA and FTA was found to provide a more complete assessment than applying either method separately [15]. Within the construction sector, falls from height have been identified as a dominant top event in Indonesian building projects [13]. Other studies have also identified falls from height, workers struck by materials, and vehicle-related incidents as major accident categories in Indonesian construction [16]. These findings confirm the recurring nature of critical construction hazards. However, they do not eliminate the need for project-specific analysis because the causes, exposure conditions, and control weaknesses associated with these accidents may differ across regions and projects.

Most Indonesian construction safety studies continue to rely on standalone methods such as Hazard Identification, Risk Assessment, and Risk Control (HIRARC) or Job Safety Analysis (JSA) [6, 17]. These methods are useful for identifying hazards and defining initial control measures, but their analytical scope is often limited. They generally do not assign differentiated weights to risk criteria, resolve the ranking ambiguities of conventional RPN, or model the causal structure of priority accidents. Consequently, the existing literature reveals a clear methodological gap. Few studies have integrated weighted risk prioritization and root cause analysis within a single framework, particularly for building construction projects in developing regional contexts such as Kupang City.

This study addresses that gap by developing an integrated OSH risk assessment framework that combines FMEA, AHP, and FTA for a building construction project in Kupang City, East Nusa Tenggara, Indonesia. The study first identifies potential hazards and associated failure modes. It then uses AHP to determine the relative importance of severity, occurrence, and detection, and applies these weights to calculate the WRPN. Finally, it uses FTA to trace the root causes of the highest-priority risks. This sequential integration is important because each method serves a distinct analytical function: FMEA structures hazard identification, AHP strengthens contextual weighting, WRPN improves risk discrimination, and FTA explains causal pathways.

The study contributes to construction safety research in three respects. First, it provides empirical evidence from Kupang City, a regional construction setting that remains underrepresented in the Indonesian OSH literature. Second, it addresses the limitations of conventional FMEA by incorporating expert-based criterion weighting. Third, it links risk prioritization with root cause analysis, allowing control recommendations to target the mechanisms that generate priority accidents rather than merely responding to their observable outcomes. The resulting framework offers a more systematic, context-sensitive, and evidence-based basis for improving OSH decision-making in regional building construction projects.

3.1 Conceptual framework

Based on the risk analysis framework shown in Figure 2, risk identification in this study was conducted systematically across all construction activities. The framework classifies construction risks into two main aspects, namely operational conditions and construction methods, which collectively influence occupational safety risks and project performance. Operational conditions include location, weather, workspace limitations, and surrounding environmental factors, while construction methods involve work execution, coordination, equipment use, and productivity pressure.

The framework also illustrates the interrelationship between contributing factors and their impacts on accidents, delays, cost overruns, and productivity decline. Therefore, this framework was used as the basis for risk identification, FMEA-AHP analysis, and the development of mitigation strategies for dominant risks in the project.

image009.png

Figure 2. Conceptual framework

3.2 Occupational safety and health risk identification

Based on the research conducted, respondent profiles were obtained. Respondent identities based on job positions were taken from the PT. Bank NTT Pension Fund Office Building Construction Project as shown in Figure 3.

In this section, risks encountered during the building construction process were identified through observation, interviews, and occupational accident history data from the project. The risks identified during the building construction process were evaluated using the FMEA method, considering the influence of severity level, occurrence frequency, and detection system. After the values were determined, the RPN was calculated for each hazard. The identified risks are presented in Table 5.

image010.png

Figure 3. Respondent profile by job position

Across the 61 identified risks, the average standard deviations of the respondents’ raw scores before rounding were 0.64 for Severity, 0.73 for Occurrence, and 0.72 for Detection. These values indicate a moderate overall dispersion in the assessments provided by the 45 respondents. However, an overall average standard deviation alone is insufficient to establish the reliability or consistency of individual risk scores because it may conceal substantial differences in respondent agreement across risk items. A low standard deviation for one risk may coexist with a high standard deviation for another, even when the overall average appears moderate. Therefore, Table 5 reports the arithmetic mean and standard deviation separately for the Severity, Occurrence, and Detection dimensions of each of the 61 risks. This item-level reporting enables a more precise evaluation of the central tendency, dispersion, and degree of respondent agreement associated with each risk score.

The mean values reported in Table 5 represent the original scores obtained from the 45 respondents, whereas the integer values of Severity, Occurrence, and Detection represent the scores generated through upward rounding for the conventional RPN calculation. This distinction is methodologically important. Although upward rounding was adopted as a conservative procedure to reduce the possibility of underestimating OSH risks, it is not statistically neutral. Upward rounding systematically assigns every non-integer mean to the next higher category, regardless of its proximity to the lower or upper integer. For example, mean values of 2.01 and 2.99 are both converted to 3, although they reflect substantially different distributions of respondent judgments. Consequently, the procedure may inflate the input scores and alter the resulting RPN and WRPN values.

Table 5. Risk identification

The potential effect of rounding is particularly important because FMEA combines Severity, Occurrence, and Detection scores multiplicatively. A one-point increase in a single parameter may produce a disproportionate change in the final risk score, especially when the other parameters are already high. The resulting ranking may therefore reflect not only the respondents’ assessments but also the mathematical consequences of the selected rounding rule. For this reason, the rounded values should be interpreted as conservative categorical representations for risk prioritization rather than exact measurements of the underlying risk magnitude.

To assess the robustness of the risk ranking, a sensitivity analysis was conducted by recalculating the WRPN values using the original unrounded mean scores for Severity, Occurrence, and Detection. These results were then compared with the WRPN values and ranking positions generated from the upward-rounded scores. The comparison showed that several risks changed their relative ranking positions, particularly those with similar WRPN values. This finding confirms that upward rounding can influence the ordering of risks when the differences between their original mean scores are small.

Despite these changes, the risks identified as the highest priorities remained unchanged under both calculation procedures. This result indicates that the identification of the most critical OSH risks was reasonably robust to the rounding assumption. Nevertheless, the stability of the highest-ranked risks should not be interpreted as evidence that rounding has no methodological consequence. The observed changes among risks with closely spaced scores demonstrate that the procedure can affect secondary prioritization and may influence decisions concerning the allocation of limited safety resources.

Accordingly, the risk-ranking results were interpreted using four complementary elements: the original arithmetic mean, the item-specific standard deviation, the upward-rounded score, and the sensitivity-analysis result. This approach provides a more transparent basis for distinguishing risks that remain consistently critical from those whose rankings depend strongly on the scoring procedure. It also prevents conservative rounding from being presented as an empirically exact representation of risk and strengthens the methodological credibility of the FMEA-AHP integration.

3.3 Failure Mode and Effects Analysis-Analytic Hierarchy Process​ analysis results

As the second step in this study, the identified risks were prioritized using the AHP method based on pairwise comparison assessments by five expert respondents (project manager, site engineer, OSH specialist, safety officer, technical manager). The AHP consistency test produced a Consistency Ratio (CR) = 0.013, which is below the required threshold (CR < 0.10), so the comparison matrix was declared consistent [11]. The weight scores for each criterion are shown in Table 6.

Table 6. Priority weight values

Note: AHP = Analytic Hierarchy Process

The AHP results indicate that Detection received the highest weight (0.623), followed by Severity (0.213) and Occurrence (0.164). Based on these weights, risks were prioritized according to their relative importance, with the Detection criterion considered by the experts to be the most important factor in OSH risk control for the studied construction project. This finding indicates that the experts emphasized the importance of early hazard detection to enable preventive actions before potential failures develop into occupational accidents. It should be noted, however, that in the FMEA rating scale, a higher Detection score indicates that a potential failure is more difficult to detect before its consequences occur. Therefore, failure modes with higher Detection ratings require greater attention because they represent lower detectability and consequently a greater need for preventive control measures. This emphasis on Detection reflects the characteristics of the project, where most activities were performed in open areas under high ambient temperatures and involved physically demanding work. Under these conditions, the ability to identify potential hazards and unsafe conditions at an early stage is essential to enable timely implementation of control measures before risks develop into occupational accidents. Therefore, the experts considered that strengthening hazard detection through more effective site supervision, routine safety inspections, continuous monitoring of workers’ compliance with personal protective equipment (PPE) requirements, and early identification of unsafe conditions would provide greater opportunities for accident prevention than relying solely on the severity of potential consequences or the likelihood of occurrence. This interpretation is consistent with the principles of proactive safety management, which emphasize the use of safety leading indicators to identify conditions that may contribute to incidents and support preventive actions before accidents occur [33]. Furthermore, this finding is consistent with the study [34], who identified weak OSH detection and monitoring systems as one of the major shortcomings of construction safety culture in Indonesia, particularly among subcontractors and specialized work units.

Although the FMEA and AHP assessments were conducted within the same construction project, the resulting risk priorities differed because the two methods served different analytical purposes. The FMEA assessment evaluated failure modes based on respondents' ratings of Severity, Occurrence, and Detection, whereas AHP was used to determine the relative importance of these criteria through pairwise comparisons performed by selected experts. Therefore, the integration of FMEA and AHP was achieved by incorporating expert-derived criterion weights into the FMEA analysis, thereby improving the reliability and robustness of the resulting risk prioritization for decision-making [10]. The WRPN calculation results are presented in Table 7. In addition to determining risk priorities based on WRPN values, risk mapping was also performed to illustrate the risk level based on the combination of Severity and Occurrence values. Each risk was plotted onto the risk matrix according to its severity level and frequency of occurrence, resulting in risk classifications of very low, low, medium, high, and very high. The risk classification used in this study followed the criteria established in the Regulation of the Minister of Public Works and Housing (PUPR) Number 10 of 2021, where risk levels are represented by different colors in the matrix: dark green for very low risk, light green for low risk, yellow for medium risk, orange for high risk, and red for very high risk.

For example, the risk of extreme fatigue in a confined excavation area (heat) obtained a Severity value of 3 and an Occurrence value of 4. Based on these values, the risk was plotted on the risk matrix at the intersection of severity level 3 and occurrence level 4. This position corresponds to the orange zone with a risk score of 16, indicating a high-risk category according to the criteria established in the Regulation of the Minister of Public Works and Housing (PUPR) Number 10 of 2021. The risk mapping results provide a visual representation of the distribution of identified risk levels and support the evaluation process in determining appropriate risk control strategies. The risk mapping results are presented in Table 8.

The two risks with the highest WRPN (3.787) in this study are extreme fatigue due to heat exposure in narrow foundation excavation areas and extreme fatigue during concrete pouring in open areas, both classified as High Risk under the risk matrix of Ministerial Regulation PUPR No. 10 of 2021, with a matrix value of 16 (Severity = 3, Occurrence = 4). This dominance of heat-induced fatigue as the top priority diverges from the majority of construction studies in Indonesia, which typically report falls from height, struck-by-material incidents, and heavy equipment accidents as dominant risks [6, 16, 17]. This distinction is attributable to Kupang City's dry tropical climate, characterized by average daily temperatures of 28–33 ℃ and high solar radiation intensity, particularly during the dry season [19], conditions that directly amplify the heat load on workers operating in open areas and confined excavation zones.

These findings are consistent with a body of literature examining the effects of heat exposure on construction worker safety. The study [35] reported that elevated ambient temperatures significantly increase the risk of injury among construction workers (RR = 1.216; 95% CI: 1.095–1.350), particularly in tasks involving intense physical activity and direct heat exposure, conditions directly analogous to the excavation and concrete pouring work examined in this study. At a broader geographic scale, heat-related injury risk tends to be substantially higher in tropical and subtropical regions than in temperate climates [36]. Ambient temperature and exposure duration have also been identified as primary determinants of heat stress, potentially causing dehydration, physical fatigue, diminished concentration, and an elevated risk of accidents [37]. Within the Indonesian construction context, the combination of extended working hours, high physical workload, and hot working environments has consistently been associated with increased worker fatigue levels [38-40]. Collectively, these findings suggest that thermal environmental factors warrant more explicit consideration in construction OSH risk assessments, particularly for projects situated in Kupang City, Indonesia.

Table 7. Risk priority and risk mapping results

Table 8. Risk level matrix

Note: Risk scores range from 1 to 25 and are classified into five categories: 1–5 = very low risk; 6–10 = low risk; 11–15 = moderate risk; 16–20 = high risk; and 21–25 = very high risk. A higher score indicates a higher priority for risk mitigation.

From a methodological standpoint, this study reinforces prior evidence on the effectiveness of integrating multi-criteria methods in risk assessment. The integration of FMEA-AHP has been shown to yield more objective risk prioritization than conventional RPN [10, 12], while the combination of FMEA-FTA-AHP enhances identification accuracy by tracing causal relationships among accident contributors [14]. This study extends those contributions by applying an integrated FMEA-AHP and FTA framework to explicitly model heat stress-induced fatigue as a failure mode, as an aspect that has largely been overlooked in risk assessments for building construction projects in Kupang City, Indonesia. This approach enables more contextually sensitive risk prioritization and root cause tracing down to the level of basic events, ultimately producing control recommendations that are both operationally grounded and adaptive to actual field conditions.

3.4 Fault Tree Analysis results

After identifying the dominant risks based on the FMEA-AHP method, the next step is to create a Fault Tree diagram that explains the causes of problems in the form of a tree diagram using standard logic symbols [22]. The resulting fault trees are presented as diagrams in Figures 4 and 5, and the FTA results for each dominant risk are further summarized in Tables 9 and 10 by explicitly presenting the corresponding top event, intermediate events, basic events, logic gates, Boolean expressions, and minimal cut sets.

image011.png

image012.png

Based on the FTA results presented in Tables 9 and 10, each top event was associated with nine basic events categorized into four intermediate events. Within the human factors category, dehydration and reduced workers’ physical condition were consistently identified as fundamental root causes contributing to both top events. Personal factors, such as physical condition and hydration, combined with environmental heat exposure, have been shown to synergistically increase construction worker fatigue [41]. Within the management factors category, excessive work duration, workload imbalance, and inadequate supervision were consistently identified as contributing basic events. Dynamic work-rest scheduling has been highlighted as an effective intervention for heat stress management, consistent with the managerial control strategies proposed in this study [37]. Periodic fatigue evaluation using both subjective and physiological approaches is also important for supporting responsive OSH management systems [42]. Within the technical factors category, repetitive manual activities, prolonged heavy equipment operation, and vibration exposure contributed to fatigue accumulation. Human factors were more dominant in this study’s FTA than in previous research, which may be attributed to dry-season conditions involving direct exposure to high temperatures and the tendency of some workers to consume energy drinks instead of water, thereby accelerating dehydration and fatigue. This finding is further supported by evidence that higher heat pressure index (WBGT) values are associated with higher fatigue levels among construction workers [40].

Table 9. Fault Tree Analysis results for extreme fatigue during excavation work

Table 10. Fault Tree Analysis results for extreme fatigue during concrete pouring due to heat exposure

3.5 Risk mitigation strategy

Mitigation strategies were developed based on FMEA-AHP and FTA findings, considering the highest Detection weighting value from AHP. Therefore, in addition to preventive measures, strategies also include early fatigue detection systems. The risk mitigation strategy is shown in Table 11.

Table 11. Risk mitigation strategy

Based on the integrated FMEA-AHP and FTA results, extreme fatigue emerged as the most critical OSH risk in soil excavation and concrete casting activities. In soil excavation, fatigue was primarily associated with a high physical workload, limited monitoring of workers’ physical condition, and prolonged heat exposure within excavation areas where air circulation and worker mobility were restricted [41]. These conditions increased physiological strain and reduced workers’ ability to maintain concentration and safe work performance.

A similar risk pattern was identified during concrete casting. Workers were required to perform continuous manual activities, including concrete distribution, compaction, and surface levelling, often under high ambient temperatures in open work areas [41, 44]. The combination of repetitive physical exertion, prolonged work duration, and heat exposure increased the likelihood of dehydration, declining stamina, slower responses, and reduced situational awareness. Extreme fatigue should therefore not be treated merely as an individual worker condition. It represents the cumulative effect of workload design, staffing adequacy, supervisory practices, work scheduling, equipment use, and environmental exposure.

The proposed mitigation measures focused on reducing the sources of physical and thermal strain rather than relying solely on workers’ ability to recognize and manage fatigue. The main measures included balancing workloads, limiting continuous work duration, establishing structured work-rest cycles, implementing job rotation, providing adequate staffing, and strengthening supervisory monitoring during high-risk activities. These measures are particularly important during excavation and casting because both activities require sustained physical effort and often allow limited flexibility once the work process has begun.

Early fatigue detection should combine behavioural, physiological, and environmental indicators. Field supervisors can identify early warning signs through direct observation, including reduced coordination, slower movement, loss of concentration, excessive sweating, and changes in worker responsiveness. Periodic physiological checks can support this process by monitoring indicators such as heart rate, body temperature, hydration status, and blood pressure. Environmental monitoring using thermal scanners or other heat-monitoring devices can also help determine when workplace temperatures exceed safe operational thresholds [41, 47].

In the longer term, digital monitoring systems may strengthen preventive control. A CCTV-based fatigue detection system could support continuous observation by identifying unusual posture, reduced movement, loss of balance, or other behavioural signs associated with fatigue. However, such a system should complement rather than replace direct supervision, adequate staffing, rest arrangements, and environmental controls. Its implementation would also require clear operating procedures, reliable detection criteria, trained personnel, and appropriate safeguards for worker privacy.

These findings underline the importance of data-supported monitoring in construction safety management. The integration of FMEA, AHP, and FTA provided a structured sequence for identifying hazards, weighting risk criteria, ranking priority risks, and tracing their underlying causes. This approach generated a more analytically complete assessment than methods that focus only on hazard identification or numerical risk ranking. More importantly, it provided a stronger basis for designing OSH interventions that respond to the actual technical, managerial, human, and environmental conditions of the project.