Datacentric System for Automatic HazOp Study

Abstract

Abstract This work aims at characterizing a system which automatically generates a list of HazOp scenarios by analyzing a set of databases which contain Process and Instrumentation Diagrams (P&IDs), its objects (equipment, piperuns, instruments, valves) and relationships between them. The scope is currently limited to process systems of FPSOs. The system can explore P&IDs through the databases that describe them, mimicking the same thinking process that human beings would execute. It divides them into nodes, and searches for possible causes of process deviations. For each cause, the system verifies which consequences could happen to P&IDs objects and, if they do happen, the system looks for detections and safeguards that could protect them, also listing interlocks. All those rules are stored as tables that objectively define the relation between causes, consequences, safeguards, carrying the engineering team experience. Based on those results, a list of recommendations is built. The result is a complete HazOp Study, including subdivided nodes, description (listing equipment, piperuns, instruments, valves, data, drawings, systems), causes, consequences, detections, safeguards, interlocks, and recommendations. Standard HazOp is very time-consuming and subject to error, as it depends on human effort and attention to details, which are not constant nor error-free. The automatic analyses are based on pre-established relationship tables of cause, consequences, and safeguards, thus more compliant to rules than the conventional HazOp. Those tables can also be customized to accommodate the operator learned experience and to allow HazOp team to add more complex scenarios. All objects from those scenarios are completely identified by tag and P&ID (typically conventional HazOp fails to be that meticulous). The interlocks related to the switches are more thoroughly searched than conventional HazOp would have (including the final element and the action). The proposed system can execute the HazOp for the entire topsides from an FPSO in about 48 hours (with no human supervision), which is a very short time when compared to requirements from a conventional HazOp. The purpose is not to completely replace standard HazOp, but to complement it. The result is standardized, fast generated, and highly detailed, contributing to higher productivity and safety.