Strategic Digitalization in Oil and Gas: A Case Study on Mixed Reality and Digital Twins

Abstract

Many organizations are developing connected worker and digital twin solutions in mixed reality as a means to quickly train new hires while simultaneously developing assistive deep learning models for quality-control mechanisms through internal documentation, diagrams, and 3D models. However, the transformation of real-life assets and processes into digital twin counterparts is a multi-step, intensive undertaking, requiring significant domain expertise and technical know-how. Therefore, we explore the architectural and technical components required to transform the most critical assets into a digital twin that yields immediate business value. Specifically, we explore the creation of digitalization architectures with re-usable components for training not only new hires but also deep neural network-based computer vision models. In this work, we present an action research case study guided by the Project Management Body of Knowledge framework. This case study was conducted in coordination with TechnipFMC, a global leader oil and gas company, on the digitalization efforts to transition into their industrial metaverse. We developed multiple architectures to bring assembly practices into a mixed reality training solution usable by trainees and editable by domain experts in real time. Further, we generate synthetic data from the same mixed reality training environments to train object detection models on industrial components and find that the photorealistic 3D models can improve mean average precision on the real-world task by +2.5 mAP.

Keywords

• Case study
• digital twin
• digitalization
• mixed reality
• need pull
• object detection