<![CDATA[Operational risks in Engineering, Procurement, and Construction (EPC) firms are interconnected and may evolve through interactions among organizational, project, and external factors. However, conventional risk assessment approaches often emphasize risk identification and prioritization without explicitly representing potential feedback relationships among risk factors. This study develops a qualitative causal framework for operational risk in Indonesian EPC firms by integrating expert-based risk-dimension validation with systems thinking and Causal Loop Diagram (CLD) modelling. The study employed a sequential two-stage approach. First, a structured literature synthesis identified 15 operational risk dimensions, which were operationalized into 45 measurement items and assessed by seven EPC experts using Content Validity Ratio (CVR), Content Validity Index (CVI), and Relative Importance Index (RII). Five core dimensions were retained: Strategic & Business Risk, Human Resource Risk, Customer & Stakeholder Risk, Environmental & Sustainability Risk, and Reputational Risk. Second, the validated dimensions were incorporated into an influence diagram and subsequently developed into a CLD. Causal directions and polarities were established through conceptual definitions, EPC operational knowledge, risk-management principles, and expert-based structural verification rather than inferred from CVR, CVI, or RII results. The resulting CLD identifies one potential reinforcing mechanism associated with risk escalation and two potential balancing mechanisms associated with governance-based and reputation-driven corrective responses. The study contributes an expert-verified qualitative causal architecture that extends operational risk assessment from variable identification toward a systems-based representation of potential feedback mechanisms. Historical pattern assessment showed directional consistency for 13 of 14 loop-specific causal relationship instances, with R1 partially supported and B1 and B2 showing consistent historical patterns.]]>
