A Conceptual Framework for Uncertainty Management in Industrial Digital Twins to Enhance Reliability, Robustness and Resilience

Digital Twin (DT) technologies are increasingly deployed in industrial systems where uncertainty management is critical to ensure their Reliability, Robustness, and Resilience R3. The resilience of the DT is consequently important, in a context where uncertainties arise from multiple sources and affect its architecture and design. For manufacturing systems, four major categories of uncertainties have been defined, namely: aleatory, epistemic, ambiguity, and interaction uncertainties. There is currently limited literature on the management of uncertainties for digital twins, and even less a structured framework for a better understanding of the uncertainties, focusing on the identification, classification, and quantification of uncertainties within the context of industrial DT engineering. In this work, a conceptual framework is proposed, based on three existing models: a conceptual model (5D: Physical Entity, Virtual Entity, Data, Service, Communication) to structure the DT architecture, ii) a DT Maturity Levels model to understand the level of complexity, and iii) a DT lifecycle model to identify critical stages. A systematic literature review using the PRISMA methodology has been conducted to map the categories of uncertainty onto these three models. Moreover, specific quantification methods have been selected and aligned with the architectural dimensions of the DT. This mapping enables a structured approach for uncertainty identification and mitigation at every stage of DT development. The proposed framework directly supports enhancing reliability by integrating uncertainty management as a core engineering principle of industrial DTs. This contribution offers a novel perspective for designing resilient DT, particularly applicable to critical industrial systems, and paves the way for future research on standardizing uncertainty management across DT engineering practices.