In Europe, the ambitious goal of targeting at least 64% of electricity production from renewables by 2050 requires some significant increase of power network ancillary services. A general extension of primary/secondary reserves is necessary to cope with the increasing penetration of stochastic renewable energies and maintain the grid vulnerability at acceptable levels. In this context, hydropower plants are called upon to play a major role due to their operational flexibility and ability to provide ancillary services. However, the provision of these services is not without consequences for the plant, as the increase of load variations and start/stop sequences enhances fatigue problems by soliciting the penstocks faster than originally expected. Given that the fatigue wear rate can be 10x higher when ancillary services are active, it is crucial to ensure the fitness-for-service of the penstocks by proper monitoring. Nevertheless, the number of sensors along the hydraulic circuit is often very limited, so that periodic stops of the plant and inspections are necessary to assess the health of the pipes. In this paper, we present how a digital twin of the power plant, namely the Hydro-Clone system, can be used to fill this gap by enabling real-time knowledge of the transient pressures throughout the water conduits. These pressures are correlated to the stress variations using either analytical formula or finite element modelling (FEM), depending on the geometry and embedding conditions of each penstock element. The validity of this approach is demonstrated by comparing the predicted stresses with measured values in the penstock of the 200 MW La Bâtiaz hydropower plant, owned by Electricité d'Emosson SA. To this end, strain gages are mounted at the bottom and top of the penstock, in front of the manifold and on the penstock protection valve. The appropriate conversion of pressure to stress at the strain gage location is derived through the analysis of FEM simulations. This work shows the benefits of using a digital twin for fatigue assessment and paves the way for real-time penstocks fatigue monitoring.