Virtual sensors augment regular sensing techniques by providing difficult-to-obtain measurements from the object of interest. By combining physics-based simulations with data-driven inference, virtual sensors provide data that cannot be obtained through physical sensors. This methodology offers enhanced predictive capabilities for case-based simulations and trend identification of phenomena embedded within structures. This research looks into the development of virtual sensing techniques for steel-reinforced concrete structures, leveraging both physical knowledge and in-situ data. To incorporate physical knowledge, a mixed-dimensional finite element model of the structure is developed using a beam-to-solid volume coupling approach. The model consists of 1D steel beam elements embedded within a 3D solid formulation for concrete. To model the uncertainty in the state of the concrete structure, random variables and random fields are defined. They model e.g. the limited knowledge about the material properties or boundary conditions. This uncertainty includes material properties, boundary conditions, and loading conditions. By sampling these random quantities, a set of simulations is created that captures the variability in the system’s response. With the set of results obtained from the simulations, a dataset is created and used to train a Gaussian process regressor. Given pseudo-sensor data from the structure, the Gaussian process regressor learns the underlying relationship between the provided sensor values and the values of interest. The trained Gaussian process regressor can then provide probabilistic predictions of the quantities of interest based on the available data at the aformentioned sensor locations and thus can be deployed in sensing applications. In summary, the virtual sensor approach presented in this work serves as a valuable tool for providing meaningful information about the state of steel-reinforced concrete structures. This addition can contribute to the protection of critical infrastructure, complementing existing structural health monitoring (SHM) systems.