*3.5. Reliability Update*

The quantified and updated uncertainties can be consistently included in the probabilistic framework to update the reliability level. The probabilistic model (1) is used to derive an annual reliability level, Δ*β*(*t*), given the updated uncertainties. This procedure is indicated in the fifth step in Figure 2, where two reliability curves (with and without using information from a digital twin) are schematically presented. The outcome of the reliability update (increase or decrease) depends on the outcome of uncertainty quantification (increased or decreased).

### *3.6. Decision Models*

Given new information from digital twins becomes available (either during operation or already in the design stage), the decision models can be updated as indicated by the last step in Figure 2. The digital twin information can be included based on Bayesian decision theory [24,55]. For existing structures, an operation and maintenance decision plan can be optimized based on an updated reliability level, for example, an updated inspection plan or lifetime reassessment. More specifically, a reliability-based inspection planning technique can be implemented [56] and some of the inspections can be removed (if any were planned during the lifetime of the structure in question) or new inspections can be included if the structural integrity is compromised. For new structures, the expected outcome of a future digital twin can be used to optimize structures already at the design stage (before the digital twin information becomes available) by the use of Bayesian pre-posterior theory [54].

#### **4. Case Study Setup**

To demonstrate an application of the proposed framework, we consider an example where information from a digital twin of an offshore wind jacket substructure is used to update the structural reliability of the substructure. The numerical models of the substructure and the turbine are described in Sections 4.1.1 and 4.1.2, followed by a description of the analyzed load case scenarios in Section 4.1.3. Based on the simulation results (in the form of stress range distributions), the structural reliability of selected joints is calculated in Section 4.2.1 by assuming a generic level of uncertainty. The results are nominal and are, in Section 5, compared with the results obtained by using digital twin information.
