Moving towards digital twins: Overview of key challenges in the process industry

Advancements in IT technologies, such as cloud computing, the Industrial Internet of Things, Big Data, Industry 4.0, and autonomous systems, are driving the adoption and digital transformation of industrial information technologies. Digital twins and modeling solutions are vital in linking physical and digital systems, creating efficient cyber-physical production systems. However, challenges hinder the digital twins' full potential in the process industry. These include finding suitable pilot projects and establishing a comprehensive life cycle management framework. This paper summarizes the main challenges and proposes possible solutions through a methodology inspired by Ops-based frameworks and model engineering. The methodology encompasses the development and life cycle management of digital twins, drawing analogies from MLOps, a framework widely used in machine learning to manage the model development process and ensure optimal performance. The novelty lies in targeting the application of simulation-based digital twins in the process industry and adapting methodologies from other fields to handle the identified challenges. The effectiveness of the proposed solutions is partially demonstrated through a real-life case study from the oil and gas industry, which showcases its adaptation and lessons learned. This research contributes to the advancement of the field of digital twins and provides valuable insights for organizations seeking to enhance their operational and decision-making processes in the era of digital transformation.