Data Management for Digital Twin in Large Capital EPCI Project

Abstract This paper presents a data management framework in the EPCI phase of FPSO projects. The data generated during the EPCI phase serves as critical upstream input for various applications in the operational phase. As advanced AI and digital twin models are increasingly adopted during operations, the integrity of the data is the foundation to meet the demands of these digital tools. This paper details our approach to handling data management and consistency, making it ready for AI and digital twin applications. In the EPCI phase, an FPSO contractor collaborates with numerous subcontractors, vendors and software applications that supply diverse types of data. This data includes structured data and unstructured data. When building a digital platform for the entire FPSO as a single source of truth, it is essential to consolidate, ensure consistency, compliance, completeness of data, and cross-check the data across the data sources linked to the digital twin. We use equipment tags as unique identifiers to link data across various sources, including 1D data (e.g., tag lists, engineering databases), 2D data (e.g., P&IDs, drawings), and 3D data (e.g., 3D models). EPC main contractor receives tag lists from multiple vendors and subcontractors. These tag lists are mostly generated manually by vendors/subcontractors which results in missing and inconsistent tags. A central tag register system is established to validate, consolidate, and correlate tags. This is then compared to our in-house developed Tag Extractor tool which automatically extracts data from PDFs using a computer vision-based AI algorithm. All output is being ingested to digital twin and have consolidated view of tag inconsistencies across mentioned sources. The developed central tagging registration system and data management framework have been successfully deployed in our recent projects. A use case study has demonstrated that this framework has been able to detect 90,000 + data inconsistencies and resolve them in early phases of MC/Commissioning and CMMS Build activities. Additionally, the KPI dashboard monitoring data consistency as the project progresses has enhanced data management efficiency.