The accurate prediction of flow separation and subsequent processes is a major challenge in computational fluid mechanics. For industrially relevant applications, the featured turbulence typically needs to be modelled at least partially to reach an affordable level of computational effort. A number of hybrid turbulence models have been developed to increase the accuracy compared to RANS but also to reduce the effort compared to LES. The presence of laminar flow regions, especially laminar boundary layers, adds further complexity since the model needs to detect those regions and become inactive. The present work combines a turbulence sensor based on instantaneous quantities with the well-known SST turbulence model to form a hybrid model which is capable of operating in URANS mode, LES-like behaviour and laminar flow. The study features the validation of the modelling approach and the demonstration of its application to typical cases where significant parts of the flow are laminar such as the process of laminar separation followed by transition into fully turbulent flow. Results of relevant test cases will we demonstrated together with the assessment of the computational effort.